Home

Software - Oracle Documentation

1. Interface SCSI Drives FC Drives SCSI Guide 3 SCSI Commands 3 SCSI ch 4 Error codes 1 HW Integration ch 10 1 HW Integration ch 6 Implementation 3 SCSI ch 1 Interpreting sense data 2 SW Integration ch 3 Messages 3 SCSI ch 2 Mode pages 3 SCSI ch 4 see the MODE SENSE command Pre execution checks 3 SCSI ch 3 Responding to Sense Keys and ASC Q 2 SW Integration ch 6 Sense Keys and ASC Q 3 SCSI ch 4 see REQUEST SENSE command Maintenance and Troubleshooting SCSI Drives FC Drives Cleaning 2 SW Integration ch 5 2 SW Integration ch 7 External drives SCSI only 1 HW Integration ch 5 n a In Libraries 1 HW Integration ch 1 In Servers SCSI only 1 HW Integration ch 4 n a In Tape Arrays SCSI only 1 HW Integration ch 3 n a Monitoring drive and tape condition 2 SW Integration ch 7 Software troubleshooting techniques 2 SW Integration ch 1 Dealing with Errors SCSI Drives FC Drives Error Codes 1 HW Integration ch 10 1 HW Integration ch 6 Handling errors 2 SW Integration ch 5 How error correction works 6 Background ch 4 Logs see the LOG SENSE command 3 SCSI ch 4 Recovering from write and read errors 2 SW Integration ch 7 Software response to error correction 2 SW Integration ch 3 Software response to logs 2 SW Integration ch 3 TapeAlert log 2 SW Integration ch 7 Ultrium
2. 6h UNIT ATTENTION Code Meaning 2800 Not ready to ready transition Action Comments A tape has been loaded successfully into the drive and is now ready to be accessed 1 The host should be polling receiving a CHECK CONDITION with sense key 2h NOT READY and additional sense of 04 O1h LUN in process of becoming ready and expecting this transition Send the console message Tape drive in process of becoming ready After the transition send the console message Tape loaded media may have changed 29 00 Power on reset or bus device reset occurred Action The drive has powered on or received a SCSI Reset signal or a Bus Device Reset message since the initiator last accessed it The action of the calling application depends on the current I O and what the operating system is expecting 1 3 As a minimum the host should re negotiate for synchronous negotiation and reconfigure the drive with any host specific operating parameters burst size bus activity limit fixed or variable mode and so on The host should then report to the console that the drive has been reset The I O can be retried if desired 2A 01 Mode parameters changed The Mode parameters for the drive have been changed by a host other than the one issuing the command UNIT ATTENTION is set for all hosts following a MODE SELECT command apart from the host that issued the command This code will only be ret
3. 6 View any returned data Exception Handling 75 Recovering from Write and Read Errors Write Recovery HP Ultrium drives ensure data integrity by performing Read After Write RAW checks while data is written to tape The drive still maintains streaming Read Recovery If all else fails the host can attempt to read through the media error See the next section Reading Through Media Errors Reading Through Media Errors In order to cope with exceptional situations the drive can read through media errors in an attempt to recover as much data from tape as possible Media errors in this case mean hard media errors from which the drive s error correction techniques are unable to recover The media errors may be caused by poor storage or tape abuse The following scenario shows how the read through facility could be used 1 Media errors of this type can be identified in response to a READ command by a sense key of MEDIUM ERROR and one of the following additional sense codes 11 00h Unrecovered read error 11 01h Read retries exhausted 31 00h Format command failed 09 00h Track following error 15 02h Positioning error detected by read of medium 2 When one of these errors occurs the drive automatically attempts to read through several groups of data on tape the exact number depends on the drive 76 Exception Handling 3 The host should continue to issue READ commands until no CHECK CONDITIONS are returned or
4. Transfer size is the amount of data transferred for a single command whether the drive is in fixed or variable block mode In both fixed and variable modes the drive works best if it receives a large amount of data for each command so a large transfer size for write commands is recommended At small block size the transfer rate is substantially degraded This is because the drive controller and the host spend too much time handling SCSI overhead instead of writing data to tape resulting in stream fails The block size at which this happens varies between drives but generally the faster the drive the larger the block size needed to stream Use 64 kilobyte transfers as a minimum Data Compression Ratio The bandwidth of the data compression engine will determine the drive s streaming capabilities based on the compression ratio of the data it is handling This is specified in the following table Compression Engine Max Streaming Product bandwidth Compression Ratio Generation 2 full height drive 120 MB s 3 1 The drive will match the throughput of any host up to the maximum native transfer rate multiplied by the current compression ratio There will not be any performance penalty for hosts that are slower than the maximum This capability is accomplished with a 64 MB buffer and use of an Adaptive Tape Speed ATS algorithm Transfer Direction Recommendation There are some noticeable performance differences b
5. ECMA EOD EOM EW EOM filemark glossary American National Standards Institute which sets standards for amongst other things SCSI and the safety of electrical devices Beginning Of Media The first point on the tape that can be accessed by the drive A mode of data transfer in write operations that facilitates tape streaming It is selected by setting the Buffered Mode Field to 7 or 2 in the SCSI MODE SELECT Parameter List header A procedure in which data is transformed by the removal of redundant information in order to reduce the number of bits required to represent the data A measure of how much compression has occurred defined as the ratio of the amount of uncompressed data to the amount of compressed data into which it is transformed The LTO DC algorithm can typically achieve a compression ratio of between 2 1 and 4 1 depending on the nature of the data A procedure in which the original data is generated from compressed data European Computer Manufacturers Association The European equivalent of ANSI End Of Data An area that signifies the end of the valid data If new data is written over a larger quantity of old data it is possible for data to exist after EOD but because it is after EOD this old data is no longer valid End Of Media format The last usable point on the tape Early Warning End Of Media A physical mark or a device computed position on the tape that tells the drive that it is approa
6. 1 HW Integration ch 7 1 HW Integration ch 4 Controller architecture 6 Background ch 4 Front Panel LEDs 1 HW Integration ch 6 1 HW Integration ch 3 Mechanism and hardware 6 Background ch 3 Specifications 4 Specs Installation and Configuration Connectors SCSI Drives FC Drives 1 HW Integration ch 7 1 HW Integration ch 4 Determining the configuration 2 SW Integration ch 2 2 SW Integration ch 2 External drives SCSI only 1 HW Integration ch 5 n a In Libraries 1 HW Integration ch 1 In Servers SCSI only 1 HW Integration ch 4 n a In Tape Arrays SCSI only 1 HW Integration ch 3 n a Modes of Usage SCSI only 1 HW Integration ch 8 n a Optimizing performance SCSI only 1 HW Integration ch 8 n a 2 SW Integration ch 4 UNIX configuration 5 UNIX Config Operation SCSI Drives FC Drives External drives SCSI only 1 HW Integration ch 5 n a In Libraries 1 HW Integration ch 1 In Servers SCSI only 1 HW Integration ch 4 n a In Tape Arrays SCSI only 1 HW Integration ch 3 n a Cartridges SCSI Drives FC Drives Cartridge Memory LTO CM 2 SW Integration ch 5 6 HW Integration ch 5 Cartridges 1 HW Integration ch 9 1 HW Integration ch 5 Features 6 HW Integration ch 5 Managing the use of cartridges 2 SW Integration ch 1 Use of cartridges 2 SW Integration ch 3
7. 32 Factors Affecting Performance Performance Factor Host Burst Rate Example Detail During the data phase of each SCSI command data is transferred to or from the drive at the host s burst rate If the host s burst rate is slow then it takes longer to transfer the data Extra time during this phase is simply added to the total command time and so it can affect the overall performance Even if the burst rate is much faster than that required to maintain streaming the total command time may prevent the commands from being issued fast enough Consider an 8 kilobyte transfer at burst rates of 8 MB s and 1 MB s The fast transfer takes 1 ms while the slow transfer takes 8 ms Since the rest of the command may only take 4 5 ms the difference of 7 ms is very significant Disk Subsystem Performance Recommendation The speed and configuration of the disks used will have a significant impact on the backup speed of the whole system Using RAID can have a significant effect on the throughput of the whole system by the use of interleaved disk reads File System Efficiency Operating systems vary in the efficiency with which they retrieve files sequentially for backup applications Most operating system development effort is put into speeding up access times within files rather than file seek times UNIX seems particularly bad at this Hardware Configuration Recommendation If the disk and tape drives are
8. 42 49 NO SENSE 41 46 NOT READY 42 47 RECOVERED ERROR 42 46 UNIT ATTENTION 43 56 VOLUME OVERFLOW 45 60 speed detecting 30 status of tape 23 support for additional LUNs 22 T Tape Capacity log 23 tape heads cleaning 15 Tape log 15 23 responding to 26 TapeAlert log 16 26 63 81 reading 73 tapemarks use of 36 tapes capacity 23 cleaning cartridge 48 faulty 53 identifying 30 length tolerance 25 monitoring condition 63 monitoring use 15 status 23 use of 14 using 23 tests read write 17 time out values 31 transfer size 13 30 effect on performance 35 troubleshooting 16 U Ultrium finding the format generation 21 supporting features 37 UNIX drivers 19 V variable length block mode 14 vendor ID 20 W Windows NT drivers 19 write errors 50 52 recovery 76 write protect 58 Index 85 86 Index
9. Condition of the Drive and Media on page 62 for more details SCSI tape drives report problems in response to a REQUEST SENSE command from the host If the backup application stores this information in a log file it becomes significantly easier to troubleshoot problems because the data can be used to pinpoint what is wrong Displaying Drive Information Troubleshooting can also be simplified by giving users the ability to look at the drive s firmware revision and information about the host bus adapter This information can be found by executing an INQUIRY command and can then be displayed or stored in a log file 16 Designing Backup Applications Drive Tests A basic read write test should be included in a backup application to check the integrity of the hardware This should also allow the user to scan the SCSI bus and to solve problems concerning the device setup and configuration Online Help Good backup applications should include online help that will allow users to search on key words and to look up error codes reported by the software or the hardware Design Goals for LTO Backup Applications suoypoiddy E Use large SCSI read write transfer sizes 64 kilobytes is recommended o 5 amp Ss w Q fey x c 5 E Incorporate data compression control and report the compression ratios achieved E Consider where to store directory information depending on the nature of the application m On
10. The Purpose of this Manual 3 Related Documents 3 Documents Specific to HP Ultrium Drives 3 Documentation Map 4 General Documents and Standardization 7 Designing Backup Applications 13 Optimizing Performance 13 Large Data Transfer Size 13 Data Compression Control 14 Non Immediate Commands 14 Managing the Use of Tapes 14 The Tape log 15 Cleaning Tape Heads 15 Monitoring Tape Use 15 TapeAlert 16 Diagnostic Logs 16 Displaying Drive Information 16 Drive Tests 17 Online Help 17 Design Goals for LTO Backup Applications 17 Configuration and Initialization 19 Operating System Drivers 19 Inquiry String Recovery 19 Product ID first 8 bytes 20 Product ID last 8 bytes 20 Product Revision Level 21 Contents 9 10 Contents Example 21 Support for Additional LUN 22 Enabling Additional LUN Support 22 Supporting Additional LUNs 22 Use of Tapes 23 LTO Cartridge Memory 23 Tape Status and Capacity 23 Finding the Remaining Capacity 24 Interpreting Log Sense Data 24 Responding to Tape Log Data 26 Load Count 26 RAW Retry Counts 26 Factors Affecting Performance 29 Ways of Optimizing Performance 29 Detecting the Drive s Speed 30 Ensuring the Recommended Minimum Transfer Sizes 30 Media Type Identification 30 Using Cartridge Memory Instead of Tape Headers 30 Tuning Performance Using the Performance Log Page 31 Time Out Values 31 Recommended Support of Log Pages 32 Factors Affecting Performance 32 Host Related Factors 32 Driv
11. The drive is ready for a cleaning periodic cleaning 1 Wait for the current operation to finish 2 Use a cleaning cartridge Check the tape drive user s manual for cleaning instructions 22 Expired C The last cleaning cartridge used in the The cleaning tape has expired cleaning media tape drive has worn out The flag is set when the tape 1 Discard the worn out cleaning drive detects a cleaning cycle cartridge was attempted but was not 2 Wait for the current operation to successful It is cleared finish internally when the next cleaning cycle is attempted m x a g o 5 aS 2 Ss Ss a 3 Use a new cleaning cartridge Exception Handling 67 No Flag Type Recommended Host Message Cause 23 Invalidcleaning C The last cleaning cartridge used in the An invalid cleaning tape type cartridge tape drive was an invalid type was used 1 Do not use this cleaning cartridge in this drive 2 Wait for the current operation to finish 3 Use a valid cleaning cartridge 24 Retension W The tape drive has requested a retension The drive is having trouble requested operation reading or writing that will be resolved by a retension cycle 25 Dual port W A redundant interface port on the tape One of the interface ports in a interface error drive has failed dual port configuration in other words Fibre Channel has failed Flags for Tape Drive Hardware Errors 26 Cooling fan W A tape drive
12. This information can be used to warn against backing up on a tape of dubious quality The Tape log also shows how many times a tape has been loaded or unloaded which gives an indication of the condition of the tape The application can then warn a user to discard the tape and use a new one if necessary Cleaning Tape Heads The Clean LED on the front of HP Ultrium drives indicates when a cleaning cartridge should be used This information can be retrieved by a SCSI REQUEST SENSE command looking at the CLN bit in the sense data When a backup application sees that the CLN bit in the REQUEST SENSE data is set it can prompt the user to clean the drive suoypoiddy go 5 amp Ss a w Q A x c 5 In an automation context the tape drive tells the automation controller that a cleaning tape needs to be used through two bits in the ACI Get Drive Status command E The Cleaning Needed bit signals deterioration in the write or read margin of the drive and indicates that a cleaning cartridge should be used as soon as possible Once the drive has been cleaned successfully the Cleaning Needed bit will be cleared E The Cleaning Required bit indicates that the drive is unable to read or write unless the drive is first cleaned so a cleaning cartridge should be used immediately Following a successful clean the Cleaning Required bit will be cleared Monitoring Tape Use Drives can report the actual amount of data that
13. a serious fault has been detected with the drive and advise them to call their technical support Log the incident in the system log Sense Keys and Codes 53 Code Meaning 1403 Manual intervention required Action Comments A tape is present in the drive but cannot be loaded or unloaded without manual intervention 40 XX Diagnostic failure on component XX Action A self test command has detected an error or a command is prohibited from execution due to failure of a previous diagnostic XX is a vendor specific code indicating the failing component see above 4400 Internal target failure Action This code is used to report hardware and firmware related hard errors that occur when the drive encounters an impossible situation see above 5100 Erase failure Action An ERASE command has failed to erase the specified area see above 5300 Media load eject failed Action A load or an eject has failed because of a problem with the tape cartridge see above 5h ILLEGAL REQUEST The following actions apply to all Additional Sense codes in this group Action 1 Terminate the current I O and return the appropriate error 2 Log the incident in the system log 3 Print out all the REQUEST SENSE data bytes and check bytes 15 through 17 to identify the location of the offending bits or bytes 4 Refer to the table of Drive Error Codes in Volume 3 The SCSI Interface
14. caused by TapeAlert This enables the software to distinguish CHECK CONDITIONs caused by actual errors and those resulting from a TapeAlert flag being set The software then reads the TapeAlert log page to discover which flags are set even for CHECK CONDITIONS caused by actual errors Note that when CHECK CONDITION results from TapeAlert the command that reported the CHECK CONDITION is not in error and will have completed successfully It follows that the software should not repeat the command E Read the TapeAlert log page using LOG SENSE at the following times Immediately after a SCSI CHECK CONDITION REQUEST SENSE cycle At the end of each tape in cases where a backup or restore spans multiple tapes If the tape is ejected the host must read the TapeAlert log page before this occurs At the end of a backup or restore It is also advisable to poll the TapeAlert log page every 60 seconds while the tape drive is idle Using Check Condition mode is recommended over polling because it guarantees that the software will be aware of any TapeAlert flag being set It is theoretically possible that TapeAlert information could be missed when polling though the higher the frequency of polling the smaller the chance is Reading the TapeAlert Log Each time the TapeAlert log page is read the application should follow this procedure 1 Read all 64 flags to discover which are set there may be more than one The
15. cycle is being performed 3 Prompt the user to wait for the cartridge to be ejected In a library the cartridge will be ejected when requested by the library or host In an internal or external single drive the cartridge will be ejected automatically 4 Prompt the user to proceed with the next application specific activity 5 Log the cleaning cycle in the system log 3700 Cleaning failure Action A cleaning operation was attempted but could not be completed for some reason Use another cleaning cartridge because the current one has expired 3A 00 Medium not present Action A medium access command has been received when no cartridge is in the drive As a minimum issue a message to the console indicating that a drive is present but no tape is loaded 48 Sense Keys and Codes Code Meaning Comments 3A 04 Media Auxiliary Memory not MAM is not ready to be accessed This is likely to be because there is no ready to access cartridge present 3E00 LUN has not self This is set during power up when it is not possible to send medium configured yet access commands to the drive because mechanism tests are being executed When the tests are complete the Additional Sense changes to 3A OOh 04 O1h or 04 02h depending on whether a cartridge was present at power on Action 1 Issue a message to the console indicating that the drive is powering up 2 Effectively poll the drive until the drive transitions to ano
16. following HP web site http www hp com support ultrium Using the NWPA model of NetWare 4 2 5 0 HP are working with Novell to provide in box support for HP s Ultrium products See the UNIX Configuration Guide for details of how to implement Ultrium support under the popular UNIX flavors Inquiry String Recovery HP Ultrium devices should not be recognized solely by the contents of their SCSI INQUIRY strings In the past hard coded recognition of Inquiry strings has meant that software support for follow on products from HP has been delayed when to all intents and purposes the new product was practically Configuration and Initialization 19 identical to the previous generation For Ultrium it is recommended that software applications key off only the first eight bytes of the Product ID field the text Ultrium The only use for the remainder of the bytes in this field is that they will be visible on screen during the boot process of PC systems As with HP s DDS products there will be very little difference between the first Ultrium drives and succeeding generations in terms of their basic SCSI characteristics they will just be bigger and faster Standard INQUIRY Page Data full height SCSI full height FC Vendor ID bytes 8 15 HP HP Product ID bytes 16 23 Ultrium Ultrium Product ID bytes 24 31 2 SCSI Z T Product Revision Level bytes 32 35 CYMV CYMV This is
17. for safe keeping are readily available for many different operating systems Unfortunately not all these applications take advantage of the advances made in tape technology over the past few years This section examines some of the characteristics that a good backup utility should include Optimizing Performance There are some fundamental things that tape management applications should implement when dealing with Ultrium drives m Use large data transfer sizes E Control and monitor data compression m Ensure directory information is safe and accurate E Maximize the use of the tape drive s internal buffering capability Each of these is discussed below For more information on optimizing performance see Factors Affecting Performance on page 29 Large Data Transfer Size Applications should use large data transfer sizes to make better of the Ultrium drive s internal buffers 64 MB A good goal to set is at least 64 kilobytes each for read or write operation E For fixed length block mode reads and writes provided the block size times the number of blocks to be transferred is at least 64 kilobytes drives will provide peak performance Small block sizes 512 bytes are Designing Backup Applications 13 acceptable so long as they are written and read in fixed length block mode in large transfers For variable length block mode reads and writes the transfer length should be at least 64 kilobytes Data Compression C
18. loads of the same tape may return different maximum capacities E At BOM remaining capacity may be less than maximum capacity This is because maximum capacity is determined when the tape is loaded Remaining capacity is determined when the request is made from the most recent tape motion data As a result the tolerances for these measurements may cause a discrepancy between the reported values 8 a 3 Use of Tapes 25 Responding to Tape Log Data Load Count Note Software should use the TapeAlert log in preference to the Tape logs to detect conditions which require the user or host to take preventative action See Monitoring the Condition of the Drive and Media on page 62 These guidelines indicate how host applications should make use of the data contained in the Tape logs during normal operation that is when tapes are not permanently write protected not constantly re formatted The console messages triggered by these criteria should clearly indicate a course of action to the end user such as the following 1 Clean the tape heads using a cleaning cartridge 2 Insert a new tape cartridge 3 Archive the data Note This only applies when non write protected cartridges are used The load count is the number of times the cartridge has been loaded into a drive and accessed Hewlett Packard recommends a maximum use for a tape of 20 000 passes over any particular area of the tape This conservativ
19. problems reading data is likely to appear with a message suggesting remedial action such as message O4h You are advised to copy any data N or message 14h clean Now Each flag is cleared to zero in the following circumstances E At power on m When specified corrective action has been taken such as using a cleaning cartridge m When the TapeAlert Log page is read Note Once cleared a flag cannot be set again until the specified clearing conditions are met So for example if the cartridge in the drive is not of data grade once flag 8 has been cleared it cannot be set again until the cartridge has been removed Designing Software to Use the TapeAlert Log 72 Exception Handling When writing software to take advantage of the ability of a drive to predict problems and actions that a user should take it is important not to exclude drives that do not support this feature For this reason the application should first check whether the TapeAlert Log page is supported by the drive and then use one of two methods to access the information E Use the MODE SELECT TapeAlert Mode page to enable Check Condition mode This means that the tape drive reports CHECK CONDITION on the next SCSI command after one or more TapeAlert flags are set When CHECK CONDITION is received the host software should behave as follows lt issues a REQUEST SENSE command Additional sense of 5D00h indicates that the CHECK CONDITION was
20. 200 GB at 2 1 compression media Factors Affecting Performance 31 Recommended Support of Log Pages Some of the media related data items on the log pages are duplicates of data that is available through the Read Attributes command using the Media Auxiliary Memory MAM access specification We recommend that you use MAM commands as the primary source for such data because this access method is portable to tape drives from other vendors that is the data is not in a vendor unique format In the long term HP intends to expose all tape usage and drive hardware usage information via the industry standard MAM format commands so it is wise to start to implement this approach now For full details of the Ultrium log pages see the LOG SENSE command in Chapter 4 Commands of the SCSI Interface Volume 3 of the HP Ultiorum Technical Reference Manual Factors Affecting Performance Further details on improving performance can be found in the How to optimize the performance of hp ultrium tape drives white paper Host Related Factors Performance Factor Host SCSI performance Detail The execution of each SCSI command involves a number of bus phases of which the data phase is only one The key phases are as follows E Intra command bus free time E Arbitration and selection E Message out Command Recommendation To achieve the highest performance the host must execute the phases that it controls quickly
21. 45 Code Sense Key Oh NO SENSE Action Console Message Interpretation These are informational positional codes The Additional Sense Codes are not generally considered errors they usually indicate some condition such as hitting a filemark The tape positional codes are mandatory for all sequential access devices This use complies with SCSI 2 so it is generic For additional sense codes see Oh NO SENSE on page 46 The host will know what fo do with this information depending upon the I O operation at the time If CHECK CONDITION occurs with this Sense Key and the Additional Sense code is not recognized by the host the software should just log the occurrence and continue It will not be considered an error The I O should have completed without an error The use of Additional Sense 8000h Cleaning Request is recommended Sense Keys and Codes 41 Code Sense Key 1h RECOVERED ERROR Console Message Interpretation These errors have been recovered by the drive The drive may report any type of recovered error Additional Sense code RECOVERED ERROR is returned is a MODE SELECT parameter is truncated or if a TapeAlert event is being reported RECOVERED ERROR is the lowest priority Sense Key it is only reported when the I O has completed with no other type of CHECK CONDITION having occurred For additional sense codes see 1h RECOVERED ERROR on page 46 None 2h NOT READY Action Cons
22. A cleaning cartridge is loaded cartridge in the drive 12 Unsupported I You have tried to load a cartridge ofa Attempted load of an format type that is not supported by this drive unsupported tape format Exception Handling 65 m x a d 5 aS 2 Ss Ss a No Flag Type Recommended Host Message Cause 13 Recoverable C The operation has failed because the The tape has snapped or mechanical tape in the drive has experienced a suffered some other mechanical cartridge failure mechanical failure failure in the drive but the tape 1 Discard the old tape can still be ejected 2 Restart the operation with a different tape 14 Unrecoverable C The operation has failed because the The tape has snapped or mechanical tape in the drive has experienced a suffered some other mechanical cartridge failure mechanical failure failure in the drive and the tape 1 Do not attempt to extract the tape cannot be ejected cartidge 2 Call the tape drive supplier s helpline 15 Memory chipin W The memory in the tape cartridge has The LTO CM chip has failed in cartridge failure failed which reduces performance Do cartridge not use the cartridge for further write operations 16 Forced eject C The operation has failed because the A manual or forced eject tape cartridge was manually de occurred while the drive was mounted while the tape drive was writing or reading actively writing or reading 17 Read only C You
23. CSI Bus Reset as follows Lock other users out cleanly Go to a minimal system single user mode Close all applications Execute a hard SCSI Bus Reset The process should not require a complete shutdown and reboot of the system Monitoring the Condition of the Drive and Media 62 Exception Handling HP Ultrium drives support TapeAlert version 3 0 in which the drive constantly monitors the condition of the mechanism and media and presents the results in a form that host software can readily use and users can easily understand You can find a general description of TapeAlert in the diagnostics section of the User s Guide Supporting TapeAlert HP has refined the TapeAlert standard to include the Predictive Failure flag If the drive sets this flag the host should display a message that the drive is about to fail and should be replaced There should be no impact on ISV software is the TapeAlert standard is already fully supported As with DDS drives the benefits of the TapeAlert system can only be fully realized if the host software supports full SNMP TapeAlert rather than just Basic TapeAlert The sending of SNMP traps in response to tape drive initiated TapeAlert flags allows network management consoles to be aware of tape drive and backup status Monitoring the Condition of the Drive and Media The TapeAlert Log page consists of 64 flags that indicate potential problems with the drive and tha
24. EOD is encountered This means the drive has reached good data and the host should concentrate on redefining the actual logical position 4 Inthe extremely unlikely situation that a large number of groups are affected by damaged media the host could still continue to issue READ commands The drive will then read through further sets of data The process is continuous providing that the host issues further READs In doing this much time is involved in recovery and a time out value of 7 minutes per READ is strongly recommended 5 Once a READ completes successfully the host should continue to issue READ commands or issue a SPACE to filemark command By reading the header after the filemark the host can establish which files are unrecoverable but more importantly can continue to recover data beyond the damaged area of tape Note It is not recommended to use the SPACE command to move over a media error In addition if a filemark is involved in the media error the SPACE command will not complete successfully This is because the filemark count will be inconsistent and the drive will start searching in the flawed area trying to find the missing filemark A filemark can only be detected on a SPACE command if the whole group containing the filemark can be read m x a d f 5 aS f 3 Ss a Exception Handling 77 78 Exception Handling ANSI BOM buffered mode compression compression ratio decompression
25. Features SCSI Drives FC Drives Adaptive Tape Speed ATS 6 Background ch 1 Autoload 1 HW Integration ch 2 Automation Control Interface ACI 1 HW Integration ch 2 6 Background ch 1 Cartridge Memory LTO CM s 1 HW Integration ch 2 2 SW Integration ch 5 6 HW Integration ch 5 Data Compression how it works 6 Background ch 5 Data Compression managing 2 SW Integration ch 5 Design principles 6 Background ch 1 OBDR and CD ROM emulation 6 Background ch 1 2 SW Integration ch 7 Performance optimization 1 HW Integration ch 8 n a 2 SW Integration ch 1 Performance factors affecting 2 SW Integration ch 4 Software design 2 SW Integration ch 1 Supporting Ultrium features 2 SW Integration ch 5 Ultrium Format 6 Background ch 2 General Documents and Standardization Small Computer System Interface SCSI 1 ANSI X3 131 1986 This is the ANSI authorized standard for SCSI implementation available through ANSI m Enhanced Small Computer System Interface SCSI 2 ANSI X3T9 2 1993 Rev 10L available through ANS Copies of General Documents can be obtained from ANSI 11 West 42nd Street New York NY 10036 8002 USA ISO CP 56 CH 1211 Geneva 20 Switzerland ECMA 114 Rue du Rh ne Tel 41 22 849 6000 CH 1204 Geneva Switzerland Global Engineering Documents 2805 McGaw Tel 800 854 7179 or 714 261 1455 Irvine CA 92714 USA Web URL http www ecma ch
26. SCSI Interface Volume 3 of the HP Ultrium Technical Reference Manual In the Performance Log page 34h parameter 04h Native drive speed gives the native speed of the drive in units of 100 KB s Drives give the value 12Ch indicating 30 MB s Ensuring the Recommended Minimum Transfer Sizes In general the larger the SCSI transfer sizes for Reads and Writes the better When operating in fixed block mode the actual block size used is largely immaterial but the size of each SCSI transfer needs to be maximized HP recommends a minimum transfer size of 64 kilobytes in order to minimize bandwidth lost to SCSI commands overheads Note that most SCSI cards working under Windows NT do not support transfer sizes greater than 64 KB without making changes to the Registry Maximum Block Size The Read Block Limits command indicates that block sizes and variable length transfer sizes are supported for values between 1 byte and 16 777 215 bytes Media Type Identification HP recommends that you use the Report Density Support command with the Media bit enabled to identify the type of media loaded in the drive The Medium Type field in the Mode Parameter Header is not used and will always be a blank field Using Cartridge Memory Instead of Tape Headers 30 For optimum performance it is also important that the host writes application tape header information to the Cartridge Memory see Cartridge Memory LTO CM on page 37 rather than t
27. Write Filemark commands Recovery from this depends on the device class and the operating system It is a generic sense key the host should be able to recover from it without knowing the additional sense code For additional sense codes see Dh VOLUME OVERFLOW on page 60 Console Message Physical End of Tape reached unable to fit remaining information on tape Additional Sense Codes Actions to Take These tables provide information about sense data so that software can know which additional sense codes can be reported under which sense keys It is important that the operating system makes all Request Sense data available to applications and in interpreted form to the end user Actions are suggested for software to use when determining the recovery action for different sense keys and additional sense code and qualifying codes ASC ASCQ It is strongly recommended that the operating system and or application use the entire ASC ASCQ data to determine the appropriate recovery action The tables are in numerical order not order of priority That is they do not suggest which sense keys should be checked first nor do they recommend priorities for the devices to report errors 1 oO gt N o A G lt N Q Ss Q O o Q o N Note When the sense additional sense code and qualifying sense keys are listed the software may look at all three keys to determine action The drive should use that exac
28. ackup Software You need backup application software that supports your Ultrium drive and tape library For the latest list of backup packages that support your Ultrium drives please contact your tape library supplier Suitable backup applications will include driver software that establishes the interface between the tape drive and the software Applications usually recognize tape drives by their manufacturers ID string rather than their model number so check the table below for the appropriate reference Drive Model ID String Generation 2 FC drive HP Ultrium 2 scs1 SCSI is not a typo Generation 2 SCSI drive HP Ultrium 2 SCSI n c me me o 3 to 5 e 3 a oO re E S oO nn Controlling Data Compression The data compression hardware in HP Ultrium drives can detect whether incoming data is already compressed and will not attempt to compress it again The drive can switch silently and with great agility between the compressing and non compressing modes thereby optimizing both compression ratio performance and data rate As this mode of behavior is embedded in the Ultrium format there is no need for host application software to switch the drive s data compression on and off Support of Data Compression Host applications may read the Data Compression Mode Page OFh to determine that the drive is capable of data compression the DCC bit is set to indicate this It is not recommend
29. ailure Action Since the function of the Test field is simply to test that an informational exception condition will produce a CHECK CONDITION and that the exception will be reported to the TapeAlert log no action is necessary Sense Keys and Codes 57 7h DATA PROTECTION Code Meaning Comments 2700 Write protect This is set if a write operation WRITE WRITE FILEMARKS or ERASE is requested for a write protected cartridge Action 1 2 3 Terminate the current I O and return the appropriate error Send a message to the console indicating that the drive has been trying to write to a write protected tape Subsequent action depends on the application 3000 Incompatible medium A write type operation could not be executed because it is not installed supported on the tape type that is loaded Action 1 2 Terminate the I O and return the appropriate error Send a message to the console indicating that writing is not allowed on the type of tape that is currently loaded Prompt the user to insert a different tape type Log the incident in the system log The calling application can retry the operation 8h BLANK CHECK Code Meaning Comments 0005 End of Data EOD detected A READ or SPACE command terminated early because it encountered EOD Action 1 Terminate the current I O and return the appropriate error to the calling application indicating that EOD has been encountered Send a conso
30. all tape drives are similar and Drivers and Logical Device Managers tend to provide connectivity based on the assumption that 80 of all SCSI tape drives behave identically A pass through mode offers the following advantages E It allows the peripheral manufacturer to provide value added diagnostics and support applications over and above those that may be shipped with a system or application E It allows System Supervisors and operators to take advantage of specific product features that are otherwise excluded because the driver or manager only caters for 80 of SCSI drives E It allows technical support people to have access to low level device specific information that otherwise is likely to be unavailable Requirements for Drivers and Logical Device Managers Drivers and Logical Device Managers must allow the user to do the following 1 Create either a 6 byte or 10 byte SCSI Command Descriptor Block 2 Allocate a write buffer or file for any data associated with the SCSI command that will be sent to the drive or Allocate a read buffer or file for any data associated with the SCSI command that will be returned by the drive 3 Link the command and data buffers Launch the command 5 The driver should use its standard CHECK CONDITION and REQUEST SENSE routines to report whether the command completed successfully or not The caller must have access to the raw REQUEST SENSE data m x a d f 5 aS ro Ss Ss a
31. ching EOM A mark written by the host It does not necessarily separate files it is up to the host to assign a meaning to the mark glossary 79 filemark count FRU hard error host load LUN LVD offline online RAW read after write SCSI 80 glossary A mark written by the host It does not necessarily separate files it is up to the host fo assign a meaning to the mark Field Replaceable Unit an assembly or group of components that is replaced in its entirety by Service Engineers when it contains a fault An uncorrectable data error The host computer system acting as controller for the drive The process in which the drive takes in an inserted cartridge and goes online Logical Unit Number by which different devices at a particular SCSI ID can be addressed individually The drive has a fixed LUN of 0 Low Voltage Differential See SCSI The drive is offline if the tape is currently unloaded or not in the drive The host has limited access and cannot perform any commands that would cause tape motion The host can however load a tape if one is inserted and can execute any diagnostic tests that do not require tape motion The drive is online when a tape is loaded The host has access to all command operations including those that access the tape set configurations and run diagnostic tests see read after write RAW improves data integrity by reading data immediately after it is written and writing
32. connection to the SCSI bus This physical arrangement is part of the SCSI 3 specification They may support either SCSI 2 or SCSI 3 protocols Wide and narrow devices can simultaneously be connected to the same bus without problem provided certain rules are followed Fast SCSI can transfer data at up to 20 MB s wide using a cable of up to 6 meters total length Ultra SCSI can transfer data at up to 40 MB s wide but the cable length cannot exceed 3 meters it is also known as Fast20 Ultra2 SCSI can transfer data at up to 80 MB s wide using a cable of up to 25 meters total length for a single device or up to 12 meters for two or more devices it is also known as Fast40 Ultra3 or Ultra160 can transfer data at up to 160 MB s wide Cable lengths are as for Ultra2 Ultra4 or Ultra320 will transfer at up to 320 MB s Ultra SCSI supports both SE and LVD interfaces In normal situations slower devices can coexist with faster devices and narrow devices can be used on the same SCSI bus as wide devices using a suitable adapter HP s Generation 1 Ultrium drives are Ultra2 wide SCSI 3 compatible devices They can be used with both LVD and SE host bus adapters Generation 2 Ultrium drives are Ultra160 wide SCSI 3 compatible single ended see SCSI TapeAlert A set of 64 flags is held in the TapeAlert log that indicate faults or predicted faults with the drive or the media By reading this log host software can inform the
33. cooling fan has failed A fan inside the drive failure mechanism or enclosure has failed 27 Power supply W A redundant power supply has failed A redundant PSU has failed failure inside the tape drive enclosure Check inside the tape drive enclosure the enclosure user s manual for ot rack subsystem instructions on replacing the failed power supply 28 Power W The tape drive power consumption is The tape drive power consuption outside the specified range consumption is outside the specified range 29 Drive W Preventive maintenance of the tape The drive requires preventive maintenance drive is required maintenance not cleaning Check the tape drive user s manual for preventive maintenance tasks or call the tape drive supplier s helpline m x a gi 5 aS 2 Ss Ss a Exception Handling 68 No Flag 30 Hardware A Type C Recommended Host Message The tape drive has a hardware fault 1 Eject the tape or magazine 2 Reset the drive 3 Restart the operation Cause The drive has a hardware fault from which it can recover through a reset m x a d f 5 aS 2 Ss Ss a 31 Hardware B C The tape drive has a hardware fault The drive has a hardware fault 1 Turn the tape drive off and then on that ent read write related or again that it can recover from through a power cycle 2 Restart the operation f The flag is set when the tape If the problem
34. ditional Sense and Qualifying Sense Keys are listed the software may look at all three keys to determine action The drive should use that exact combination to report that particular error For example a drive will report that it is not ready when there is no cartridge present by setting Sense Key NOT READY with Additional Qualifying Sense Keys 04 02 No other combination of Additional Qualifying Sense Keys may be used to report that particular condition Exception Handling These pages cover methods of dealing with certain error conditions and exceptional circumstances They include the following E A suggested Escalation Procedure to follow when exceptions occur E How to make the most of the TapeAlert facility see Supporting TapeAlert on page 63 E How to respond to the Clean LED lit on the front panel when there are an excessive number of retries or error corrections E The need for a pass through mode which can return information or commands specific to a vendor s product m How drives recover from read and write errors How to use the drive s ability to read through media errors so that as much data can be retrieved from a badly damaged tape as possible Typical Escalation Procedure For exception handling there needs to be a well defined escalation path through which the calling application user operator or System Supervisor may take increasingly drastic action to clear any product related
35. e Related Factors 34 Format Related Factors 36 Supporting Ultrium Features 37 Cartridge Memory LTO CM 37 Cleaning 37 Resetting Drives 38 Backup Software 39 Drive Model ID String 39 Controlling Data Compression 39 Support of Data Compression 39 Other Mode Page Information 40 Buffer Size at EW EOM 40 Partition Size 40 Sense Keys and Codes 41 Sense Keys Actions to Take 41 Additional Sense Codes Actions to Take 45 Oh NO SENSE 46 Jh RECOVERED ERROR 46 2h NOT READY 47 3h MEDIUM ERROR 49 Ah HW ERROR 53 5h ILLEGAL REQUEST 54 6h UNIT ATTENTION 56 7h DATA PROTECTION 58 8h BLANK CHECK 58 Bh ABORTED COMMAND 59 Dh VOLUME OVERFLOW 60 Exception Handling 61 Typical Escalation Procedure 61 Monitoring the Condition of the Drive and Media 62 Supporting TapeAlert 63 Monitoring the Condition of the Drive and Media 63 Flags 63 Flags for Tape Autoloader Errors 70 Designing Software to Use the TapeAlert Log 72 Reading the TapeAlert Log 73 Responding to the Clean LED 74 Providing Pass Through Mode 75 Requirements for Drivers and Logical Device Managers 75 Recovering from Write and Read Errors 76 Write Recovery 76 Read Recovery 76 Reading Through Media Errors 76 Contents 11 a e 3 o 3 un Glossary 79 Index 83 12 Contents Designing Backup Applications In today s computer market software applications that use tape drives to copy the information from a computer s hard disk
36. e changes to 3A OOh or Y 04 02h 2 ls a A 3 Instruct the user what to do based on the application and the 2 previous sequence of commands g Bs 4 Depending on the application the software may terminate the 9 current I O F 0401 LUN in process of A media access command has been received while a load or becoming ready unload is already occurring with immediate report on or initiated through the front panel Action 1 Effectively poll the drive by re sending the command until the media is loaded when UNIT ATTENTION with Additional Sense of 28 OOh will be set At this point the command can be executed 2 Report to the console that the drive is in the process of loading tape Sense Keys and Codes 47 Code Meaning 0402 LUN not ready initializing command required Action Comments A cartridge is present in the drive but is not logically loaded 1 The software must issue a LOAD before media access is permitted 2 Issue a message to the console to request a user initiated load or to indicate drive status or both 0410 Media Auxiliary Memory is The Media Auxiliary Memory is not accessible not accessible 3003 Cleaning cartridge installed Action A medium access command has been sent to the drive while a cleaning cartridge was loaded 1 Terminate the current I O and return the appropriate error 2 Send a message to the console indicating that a cleaning cartridge is in the drive and a cleaning
37. e estimate is also influenced by the quality of the application and the driver software in being able to maintain streaming thereby preventing repositioning over the same area of tape without data being transferred RAW Retry Counts 26 Use of Tapes Data is read immediately after being written to tape to establish that it has been written correctly Increases in RAW retries can be due to four factors E Deterioration in the media m Dirty heads E Drive malfunction E The operating environment Corrective Action The recommended criteria for corrective action are as follows RAW Retries gt 5 Total groups written When using tapes without write protection use the Total count The corrective action should be as follows 1 Use another tape and for a write operation try repeating the write For a read operation try reading data from the tape 2 See whether the current RAW value is within the recommended limit 3 If the values are now within the limit you can assume that the original tape is nearing the end of its useful life Proceed as follows Fora write operation discard the tape and use a new one For a read operation transfer the data to a new tape 4 If the value is still outside the limit clean the tape heads with a cleaning cartridge and try repeating the operation with the original tape 8 a 3 Use of Tapes 27 28 Use of Tapes Factors Affecting Performance A This chapter con
38. ed that the host attempts to modify either the Data Compression Enable DCE or Data Decompression Enable DDE bits Similarly the Select Data Compression Algorithm SDCA field in the Device Configuration Page page 10h is set to 1 on and it is not recommended that this is changed by application software Supporting Ultrium Features 39 Other Mode Page Information Buffer Size at EW EOM The Buffer Size at EW EOM field in the Device Configuration Mode page 10h is set to zero as with HP DDS tape drives and other tape technologies HP Ultrium tape drives automatically allow sufficient space between EW EOM and physical EOT to satisfy backup applications Partition Size The LTO Ultrium format only supports a single partition There is no Medium Partition Mode page 11h In the Device Configuration Mode page 10h the Change Active Partition CAP flag and the Active Partition field should both be zero since multiple partitions are not supported 40 Supporting Ultrium Features Sense Keys and Codes Sense Keys Actions to Take Ultrium drives follow the ANSI definition of sense keys The following table explains how the drives interpret sense key descriptions As sense keys and additional sense codes are intended to be hierarchical errors the table recommends action for the host when a particular sense key is reported For more detailed recovery actions see Additional Sense Codes Actions to Take on page
39. etween reads and writes caused by the extra device CPU time needed by the drive to read frames The drive is less likely to stream read small transfer sizes than it will when writing transfers of the same sizes Raw Tape Data Rate This is the maximum rate at which data can physically be written to tape This is 30 MB s Factors Affecting Performance 35 upwoj d n Q Tay 2 Ke 7 ES o 2 3 Q Performance Factor Detail Recommendation Select the maximum data transfer rate for the type of application used Format Related Factors Performance Factor Detail Tapemarks Tape marks filemarks have many different uses to give a logical structure to data on a tape The SCSI Standard specifies certain actions that the drive must take when it is told to write a filemark If the drive is told to write a filemark when the Immediate bit is not set the standard insists that the drive must flush all data to tape If used unnecessarily this will adversely affect performance and waste tape capacity 36 Factors Affecting Performance Supporting Ultrium Features 5 This section covers the following features of HP Ultrium drives LTO Cartridge Memory Automation Control Interface ACI Controlling data compression Other Mode page information the Buffer Size at EW EOM field and partition size Use in tape libraries Cartridge Memory LTO CM LTO Contactless Memory LTO CM is EEPROM memory that is embedded i
40. faults An escalation procedure is important to allow local recovery where possible and fo avoid the unnecessary replacement of peripheral devices A typical escalation procedure is as follows 1 Retrieve fault information Specific fault information is retrieved from the following sources INQUIRY data such as firmware revisions REQUEST SENSE data such as Additional Sense Codes and Drive Error Codes Exception Handling 61 MODE SENSE data such as data on the current configuration LOG SENSE data stored in the drive s logs 2 Inform the user The system gives the user helpful advice by attempting to decode the returned information and also allows the user access to the raw data 3 Allow the user to try recovery For any fault the system allows the user to use simple recovery commands such as REWIND or LOAD UNLOAD 4 Allow the user to reset devices If these actions fail including repeated attempts to retry the operation the user should be able to reset devices on the SCSI bus selectively through the use of a BUS DEVICE RESET message This message is specific only to the device to which it is addressed and will not disturb other devices on the SCSI bus 5 Allow the System Supervisor or support person to perform a controlled hard SCSI Bus Reset If the BUS DEVICE RESET message fails to clear the problem the System Supervisor or technical support person should be able to perform a controlled hard S
41. has been written to the tape and the amount of available space on the tape From this information applications for Ultrium drives can be designed to calculate the percentage of tape used and give the user feedback on the actual progress of the backup operation This is a significant improvement over other technologies such as Designing Backup Applications 15 TapeAlert Diagnostic Logs DC6000 QIC products that require the application to estimate what is going on See Tape Capacity Log Page under the LOG SENSE command in Chapter 3 of The SCSI Interface Volume 3 of the HP Ultrium Technical Reference Manual for more information While the reliability of tape products and applications is getting better all the time problems do still occur There are some very simple techniques that could be incorporated by application developers to simplify the process that a user must go through to resolve problems For additional information see Exception Handling on page 61 The TapeAlert facility in HP Ultrium drives allows applications to help avoid trouble by prompting the user to take remedial action or in some cases through the application automatically performing remedial actions itself For example if the drive is experiencing trouble writing the software can prompt the user to clean the heads or if there are several drives or an autoloader automatically clean the heads without involving the user See Monitoring the
42. have loaded a cartridge of a type A write command has been format that is read only in this drive The attempted to a tape whose cartridge will appear as write protected format is read only in this drive 18 Tape directory W The tape directory on the cartridge has The drive was powered down corrupted on been corrupted File search with a tape loaded or a load performance will be degraded The tape permanent error prevented the directory can be rebuilt by reading all tape directory being updated the data on the cartridge m x a d 5 ae 2 Ss Ss a Exception Handling 66 No Flag Type Recommended Host Message Cause 19 Nearing media The tape cartridge is nearing the end of The tape may have exceeded its life its calculated life It is recommended that specified number of passes you 1 Use another tape cartridge for your next backup 2 Store this tape cartridge in a safe place in case you need to restore data from it Flags for Cleaning Management 20 Clean now C_ The tape drive needs cleaning The tape drive has detected that it needs cleaning The flag is cleared internally when the drive is cleaned successfully E If the operation has stopped eject the tape and clean the drive E f the operation has not stopped wait for it to finish and then clean the drive Check the tape drive user s manual for cleaning instructions 21 Clean periodic W The tape drive is due for routine
43. he current I O and return the appropriate error 2 Send a message to the console indicating that there is a problem with the format of the tape in the drive 3 Prompt the user to eject the media and insert a valid Ultrium tape 4 Log the incident in the system log Sense Keys and Codes 51 1 oO gt N o A G lt N Q Ss Q O Q oO N Code Meaning Comments 3B 00 Sequential positioning error The drive has failed to read groups off tape There are two possibilities Action 4 5 The current command such as READ SPACE REWIND or WRITE failed to complete successfully The logical position has been lost The tape has been positioned on the other side of the bad groups The host must determine if it is worth continuing which presents a danger of missing EOD and running into old data The host must also decide if the data is old or current Attempt to recover by executing a REWIND command to return to a known position such as BOT Space to the position of the last known successful command and retry the failing command If this is unsuccessful terminate the current I O and return the appropriate error Prompt the user to back up the data to a new cartridge and discard the old one Log the incident in the system log 5000 Write append error Action A write type command failed because the point at which to append data was unreadable This was probably caused by a powerfai
44. idges Chapter 9 of the Hardware Integration Guide Volume 1 of the HP Ultrium Technical Refernence Manual also contain information on Ultrium cartridges Timing considerations are discussed in Time Out Values on page 31 LTO Cartridge Memory Cartridge Memory has been added to the LTO cartridge for the following reasons E It speeds up load and unload times by removing the need to read system areas E t speeds up movement around tape by storing the tape directory physical to logical mapping E It increases tape reliability because fewer tape passes are needed E It stores diagnostic and log information for tracking purposes Most of these uses are invisible to applications and handled internally by the drive There is potential for some other applications to use the Application Specific Data area This is being investigated For more details see Using Cartridge Memory in Using Special Features in Libraries Chapter 2 of the Hardware Integration Guide Volume 1 of the HP Ultrium Technical Refernence Manual Tape Status and Capacity Following autoload or a load command the software can determine the state of the tape and its capacity from the Tape Usage Log and Tape Capacity Log Use of Tapes 23 Units General Maximum Capacity Remaining Capacity 24 Use of Tapes pages retrieved through the LOG SENSE command The information can also be invoked as a console operation at any time to f
45. ind the status and condition of the media Tape capacity figures can be used for two purposes To give an application or user an indication of whether the tape has enough capacity for a proposed backup When using data compression however this is of little value since the compression factor cannot be predicted accurately Periodically during a backup to give an approximation of the amount of tape left Caution An application should not use the capacity reported in the Tape Capacity log to fix the backup size This will result in permanent capacity truncation that could represent a significant percentage of the available capacity Finding the Remaining Capacity Examine the Tape Capacity Log to estimate the effective remaining capacity of the tape data compression factors are not considered Interpreting Log Sense Data The following points affect the values returned in the data Capacities are given in kilobytes of user data If data compression is used the capacities are specified as though the drive is in pass through mode The data compression factor is not considered System log area vendor group and EOD areas are not included in capacities specified In other words values are conservative An allowance for read after write retries is made Maximum capacity values are only valid when the tape has completed a load sequence If an immediate mode LOAD is made LOAD SENSE will not return valid information until the
46. inues to fail with the same sense key The specific retry or recovery strategy depends on the operating system Illegal SCSI command requested 6h UNIT ATTENTION The operating conditions of the drive have been changed in some manner that the host should be aware of For example the drive may have gone online or been reset the Mode parameters may have been changed a second host may have changed the drive s operating conditions and so on For additional sense codes see 6h UNIT ATTENTION on page 56 Also see the Unit Attention pre execution check Sense Keys and Codes 43 1 gt N ko A oO lt N Q Ss Q O o Q oO N Code Sense Key Action Console Message Interpretation Recovery depends on the device class and the additional sense code For most codes the software should re initialize the drive The driver should repeat the command If additional sense code 290Xh the driver should renegotiate SDTR and reconfigure the drive through MODE SELECT Tape drive operating conditions may have changed 7h DATA PROTECTION Action Console Message This is an error if the I O operation is attempting to access the media in some manner and failing because the media is write protected For additional sense codes see 7h DATA PROTECTION on page 58 Also see the Media Write pre execution check Depends on the operating system If this sense key occurs with other than an addit
47. ional sense code of 2700h the software should log the error terminate the I O and then send an error operating system specific to the calling application Data on the tape will not have been altered Media write protected 8h BLANK CHECK Console Message An attempt was made to read unwritten media Recovery from this sense key depends on the operating system It may be regarded as an error because more data was expected by the host or it may be an expected condition For additional sense codes see 8h BLANK CHECK on page 58 End of Data encountered Bh ABORTED COMMAND Action Console Message The drive has given up on a command This is normally due to a problem related to the SCSI bus For example it is reported if a target or LUN receives a second command from the same host before the previous command from that host has completed For additional sense codes see Bh ABORTED COMMAND on page 59 Recovery depends on the additional sense code and the operating system In some cases the host may want to retry the current I O If the additional sense code is 4EOOh overlapped commands attempted the host may not want to retry the current I O because the previous I O will not have been completed SCSI protocol problem 44 Sense Keys and Codes Code Sense Key Interpretation Dh VOLUME Data could not be written because of a lack of remaining space on the OVERFLOW tape See the Write and
48. l end of the tape was encountered The EOM flag in the sense data will be set Action see above 82 82 Drive requires cleaning The drive has detected that the heads need to be cleaned to maintain good operation Action Optionally log the occurrence for information It will not be considered an error and the software will continue 1Th RECOVERED ERROR Note Reporting of recovered errors defaults to OFF Action In all cases action depends on the device class and operating system 46 Sense Keys and Codes Code Meaning Comments 3700 Rounded parameter The drive needs to round off the value of a parameter sent by MODE SELECT because it cannot store it to the degree of accuracy sent by the command 5D 00 Failure prediction Failure Prediction thresholds have been exceeded indicating that a threshold failure may soon occur 2h NOT READY Code Meaning Comments 0400 LUN Not Ready no cause This is set if a LOAD UNLOAD command is received while a load or to report unload is already occurring with immediate report on or initiated through the front panel It is present for the duration of the unload or eject after which the additional sense changes to 3A 00h medium not present or 0402h logical unit not ready initializing command required Action 1 Issue a message to the console stating that the tape is currently being unloaded from the drive 2 Poll waiting until the Additional Sens
49. l or SCSI bus reset preventing the drive from completing a write operation properly and appending an EOD 1 2 3 Terminate the current I O and return the appropriate error Tell the user that the append point is unreadable Advise the user to back up the data to new media and reformat the failing tape Log the incident in the system log 52 Sense Keys and Codes Code Meaning Comments 5200 Cartridge fault A command could not be completed because of a fault with the tape cartridge Action 1 Terminate the current I O and return the appropriate error 2 Tell the user that a serious fault has been detected with the tape cartridge 3 Advise the user to discard this cartridge and select a new one 4 Log the incident in the system log 5300 Media load eject failed A load or eject has failed Action 1 Terminate the current I O and return the appropriate error 2 Inform the user that a serious fault has been detected with the media 3 Advise the user to discard this cartridge and select a new one 4 Log the incident in the system log Ah HW ERROR The following actions apply to most Additional Sense Codes for HW ERROR Sense Key Action 1 Terminate the current I O and return the appropriate error 7 e gt n o A G lt N Q Ss Q O Q oO N The software should disable all further transactions to the drive and mark the drive as bad The software should tell the user that
50. le message saying that EOD has been encountered Recovery depends on the calling application and what was expected 58 Sense Keys and Codes Code Meaning 1403 End of Data not found Action Comments A read type operation failed because a format violation related to a missing EOD data set The most likely cause is a tape with corrupt format perhaps from a powerfail when the tape was being written 1 Terminate the current I O and return the appropriate error 2 Send a message to the console indicating that EOD could not be found because the tape has corrupt format 3 Prompt the user to back up the data to another tape and discard the current one 4 Log the incident in the system log Bh ABORTED COMMAND The following action applies to all codes in this group except 00 OOh and 44 00h Action 1 Terminate the current I O and return the appropriate error 2 Log the incident in the system log 3 Send the console message SCSI command aborted low level failure on SCSI bus Code Meaning 3D 00 Invalid bits in IDENTIFY message Comments The drive has received an illegal IDENTIFY message at the start of a command 4300 Message error The number of parity errors or problems on inbound or outbound messages has exceeded the limit Note that it is likely that a Bus Release will occur if excessive parity errors are detected since the drive will fail to complete the Status phase and Command C
51. ly use immediate WRITE FILEMARK commands but avoid the using other commands in immediate mode m Use Tape Log information to measure tape quality before backing up starts m Use the TapeAlert log to prompt the user to take remedial action to avoid problems E Use cleaning required indicators in the software to prompt the user to clean the drive heads E Allow users to set custom cleaning schedules E Use log files to store Inquiry and Sense Key Error Code information about error conditions E Allow users to access drive firmware revision and HBA characteristic information E Include the capability to download firmware E Incorporate simple diagnostic capabilities such as Write Read tests and SCSI bus scanning E Incorporate online help Designing Backup Applications 17 18 Designing Backup Applications Configuration and 2 Initialization This section covers the following topics E Operating System drivers E Inquiry string recovery finding information about the drive through the Inquiry command m Additional LUN support for operation with an autochanger device E Fibre Channel support Operating System Drivers Windows NT NetWare UNIX HP have a proprietary performance optimized driver for Windows NT 4 Windows 2000 and Win NET It is intended that the driver will be freely licenced to any software partner that requires it For the latest driver support for HP tape drives please visit the
52. mat corrupt 51 incompatible 51 front panel LEDs 74 FRU 80 G generation finding 21 H hard error 80 help 17 host 80 l initialization 19 INQUIRY command 16 inquiry string recovery 19 L LEDs Clean 74 load count 26 failed 53 54 loading 80 LOG SENSE data 24 logs Diagnostic 16 performance 31 supporting 32 Tape 15 23 26 Tape Capacity 23 TapeAlert 16 26 63 81 LTO CM 23 37 using instead of headers 30 LUNs 80 not ready 47 support for additional 22 M MAM write error 50 51 media errors reading through 76 media see tapes memory cartridge 23 microcode 57 mode fixed length block 13 pass through 75 variable length block 14 mode parameters 56 monitoring condition 16 62 monitoring tape use 15 N NetWare drivers 19 non immediate commands 14 O offline 80 online 80 online help 17 operating system drivers 19 optimizing performance 13 29 P parity error 59 partition size 40 pass through mode 75 performance factors 29 drive related 34 format related 36 host related 32 performance log 31 problems predicting 16 62 product ID 20 product revision level 20 21 R RAW 76 80 retries 74 retry counts 26 read errors 50 reading through 76 recovery 76 reading the TapeAlert log 73 remaining capacity 24 revision level 20 21 S SCSI 3 80 sense keys ABORTED COMMAND 44 59 BLANK CHECK 44 58 DATA PROTECTION 44 58 HW ERROR 43 53 ILLEGAL REQUEST 43 54 MEDIUM ERROR
53. may have a hardware required fault Run extended diagnostics to verify fault that may be identified by and diagnose the problem extended diagnostics using a Check the tape drive user s manual for SEND DIAGNOSTIC command instructions on running extended diagnostic tests Flags for Tape Autoloader Errors Flags 40 49 are not currently used 50 Lost statistics W Media statistics have been lost at some The drive or library has been time in the past powered on with a tape loaded 51 Tape directory W The tape directory on the tape cartridge An error has occurred invalid at just unloaded has been corrupted File preventing the tape directory unload search performance will be degraded being updated on unload The tape directory can be rebuilt by reading all the data 52 Tape system C The tape just unloaded could not write Write errors occurred while area write its system area successfully writing the system area on failure 1 Copy the data to another tape unload cartridge 2 Discard the old cartridge 53 Tape system C The tape system area could not be read Read errors occurred while area read successfully at load time reading the system area on failure load Copy the data to another tape cartridge 70 Exception Handling No Flag Type Recommended Host Message Cause 54 Nostartofdata C The start of data could not be found on The tape has been damaged the tape 1 Check that you are using
54. n every LTO Ultrium tape cartridge It is non volatile and is contactless in that it is read by RF coupling rather than electrical contact Further Information For general information about LTO CM see LTO Cartridge Memory in Chapter 5 Cartridges in Background to Ultrium Drives Volume 6 of the Ultrium Technical Reference Manual For suggestions of how to make use of cartridge memory in libraries see Using Cartridge Memory in Using Special Features in Libraries Chapter 2 of the Hardware Integration Guide Volume 1 of the HP Ultrium Technical Reference Manual Cleaning The tape drive tells the automation controller that a cleaning tape needs to be used through two bits in the ACI Get Drive Status command The Cleaning Supporting Ultrium Features 37 Needed bit indicates deterioration in the write or read margin of the drive and hence it is recommended that a cleaning cartridge be inserted into the drive at the earliest opportunity Following a successful clean the Cleaning Needed bit will be cleared The Cleaning Required bit indicates that the drive is unable to read or write unless the drive is first cleaned with a cleaning tape It is recommended that a cleaning cartridge be loaded into the drive immediately Following a successful clean the Cleaning Required bit will be cleared Cleaning cartridges can be used for 50 cleanings Resetting Drives The tape drive can be reset by the automation contr
55. nd discard the current tape 50 Sense Keys and Codes Code Meaning Comments 11 12 Media Auxiliary Memory read An error has occurred while attempting to write to MAM The cartridge error should not be used for further backups but should be able to be used for restoring data 14 00 Recorded entity not found Action A SPACE or LOCATE command failed because of the drive could not find the target of the operation because of a format violation 30 01 Cannot read media unknown format Action A LOCATE or SPACE command has found the tape is in a format not supported by the drive 1 Terminate the current I O and return the appropriate error 2 Send a message to the console indicating that the tape is in a format not supported by the drive 3 Prompt the user to eject the cartridge and insert a valid one 4 Log the incident in the system log 30 02 Cannot read media incompatible format Action A READ command could not be completed because the logical format is not correct 1 Terminate the current I O and return the appropriate error 2 Send a message to the console indicating that the tape is wrongly formatted 3 Prompt the user to eject the cartridge and insert a valid one 4 Log the incident in the system log 31 00 Medium format corrupted Action READ or SPACE has tried to read data that is in a format that is recognized but which is not valid 1 Terminate t
56. ng products are covered Capacities are when the drive is using data compression with a compression ratio of 2 1 where applicable E HP Ultrium Generation 2 Full Height SCSI Internal Drive E HP Ultrium Generation 2 Full Height FC Internal Drive Note Throughout this manual frequent reference is made to SCSI commands For more information on SCSI commands for HP Ultrium drives see volume 3 The SCSI Interface of the HP Ultrium Technical Reference Manual set Ordering details are given below Related Documents The following documents provide additional information Documents Specific to HP Ultrium Drives Hardware Integration Guide volume 1 of the HP Ultrium Technical Reference Manual E The SCSI Interface volume 3 of the HP Ultrium Technical Reference Manual E Specifications volume 4 of the HP Ultrium Technical Reference Manual E HP Ultrium Configuration Guide volume 5 of the HP Ultrium Technical Reference Manual Background to Ultrium Drives volume 6 of the HP Ultrium Technical Reference Manual Please contact your HP supplier for copies E The features and benefits of HP Ultrium drives are discussed in the HP Ultrium Technology White Paper E For a general backgrounder on LTO technology and licensing go to http www 1lto technology com Documentation Map The following will help you locate information in the 6 volume Technical Reference Manual Drives general Connectors SCSI Drives FC Drives
57. not a typo Product ID first 8 bytes Ultrium This will be the same for all HP Ultrium products regardless of generation or model Product ID last 8 bytes Ist byte Generation identifier Ula 2 Generation 2 400 GB at 2 1 compression 2nd byte Hyphen separator ASCII 2Dh 3rd 6th scst Parallel SCSI interface and Generation 2 FC drives a n bytes Fc Fibre Channel interface although not for Generation 2 drives 20 Configuration and Initialization Product Revision Level Ist byte Product codename ID F Generation 2 full height drive 2nd byte Firmware year o 9 2000 2009 3rd byte Firmware month au a 1 B January December Ath byte Firmware variant Mow D Standard distribution firmware Example If new drive families or variants support features that are not available in previous generation products you can detect the existence of these features through the SCSI Mode Sense and Log Sense commands Exact details will become available as new products are defined There is no need to limit driver or application connectivity to a single HP Ultrium product type To determine the drive technology family Examine only the first eight bytes of the Product ID field the text u1trium To determine the Ultrium format generation Use one of the following two methods of which the second is preferred E Examine the single characte
58. o the actual tape This allows cartridges to load and unload quickly and prevents excessive media wear at the beginning of the tape As the access method to Cartridge Memory data is an open standard it also permits other software systems to identify alien media positively in shared storage environments Factors Affecting Performance Tuning Performance Using the Performance Log Page The Performance Log page 34h contains data that should allow application software to tune the data rate being sent to the drive dynamically SIUDULIO I3d n Q Tay 2 Ke Z z o 2 3 Q Time Out Values SCSI Command Recommended Time Out Value Load 10 minutes Unload 10 minutes Rewind full tape length 10 minutes Space Locate 20 minutes Erase long 5 hours Erase short 5 minutes Write Write Filemarks 5 minutes Read 5 minutes Read Write Attribute MAM with 1 minutes 1 KB of attribute data Non tape movement such as TEST 1 minutes UNIT READY INGUIRY Notes E These values are for a single SCSI command in non Immediate mode As most commands will be sent in Immediate mode status will be received by the host typically within 20 ms In such cases the time out given indicates when the drive will have completed the operation and be ready for the next tape movement command All of these values may be subject to change There is no retension facility Where relevant the figures above apply to
59. of the Ultrium Technical Reference Manual to understand why the current values were rejected 5 This is likely to be an application fault Send a message to the console Illegal SCSI request to tape drive 54 Sense Keys and Codes Code 1A 00 Meaning Parameter list length error Comments A MODE SELECT parameter list sent to the drive contains one of the following m An incomplete Mode Parameter header must be 4 bytes E An incomplete Mode Block Descriptor must be O or 8 bytes E An incomplete Mode page 20 00 Invalid command opcode The drive does not recognize the opcode of the command it has received 24 00 Invalid field in CDB The drive has detected an invalid field in a command descriptor block 25 00 LUN not supported The command was addressed to a logical unit number that does not exist 26 00 Invalid field in Parameter List The drive detected an invalid field among the command parameters sent during the data phase 26 01 Parameter not supported A request for an invalid page number has been sent 53 02 Medium removal prevented An UNLOAD command failed to eject the tape because medium removal has been prevented 82 83 Bad microcode detected The data transferred to the drive during a firmware upgrade is corrupt or incompatible with the drive hardware Sense Keys and Codes 55 17 oO gt N o A G lt N Q Ss Q O Q oO N
60. ole Message This Sense Key generally means the host will have to wait for the drive to become READY Media access is not possible Click here for additional sense codes Also see the Media Access pre execution check The host needs to take one of the following actions E Wait until the drive becomes available E Issue some type of initializing command E Perhaps instruct the user to put the drive online The handling of this Sense Key will depend upon the host s operating system and the Additional Sense code Drive not ready media access not possible 3h MEDIUM ERROR Action 42 Sense Keys and Codes This sense key indicates a failure that is probably due to a problem with the tape The I O did not complete The O may have been partially attempted and data on the tape may have been altered The drive will have retried an optimal number of times before reporting this sense key For additional sense codes see 3h MEDIUM ERROR on page 49 Recovery depends on the operating system or application At the very least whatever the additional sense code the software should log the error terminate I O to the drive and pass the appropriate error to the calling application On read the driver should discriminate between hard read failures caused by the media and those resulting from an inability to decompress data Code Sense Key Console Message Interpretation On write Additional Sense OCOOh Write to ta
61. oller via the ACI Reset command or by pulling the ACI_RST_L line low see Rear Panel and Connectors Chapter 7 of the Hardware Integration Guide Volume 1 of the HP Ultrium Technical Reference Manual Two levels of reset via the ACI interface are provided namely ACI Reset and Drive Reset The former resets the ACI port and all SCSI parallel Fibre Channel ports The latter is equivalent to a power on reset Either reset method will interrupt the interface between the drive and host with the Drive Reset potentially resulting in no End of Data written to tape It is therefore strongly recommended that an ACI reset command is not sent unless all other recovery methods have failed Note that certain ACI commands see Table 5 can be queued behind outstanding SCSI commands giving the impression that the drive has stopped responding over the ACI bus All command packages will be still be ack ed even though the command will be queued Following an ACI Reset command with reset control set to Drive Reset or after pulling the ACI_RST_L line low the drive will behave as if it has powered up and will go off bus and lose all ACI configurations A SCSI interface reset will not affect the ACI interface Note that following an upgrade of the drive firmware either via tape or SCSI the drive will be reset as if it had been powered up The implementation details are beyond the scope of this document currently 38 Supporting Ultrium Features B
62. omplete message 4500 Select or reselect failure The drive has failed to reselect a host to complete an operation CHECK CONDITION state is assumed to exist even though it was never reported 4700 SCSI parity error The drive has detected a parity error in an unexpected SCSI state It should never do this so this error should be treated as a firmware error Sense Keys and Codes 59 17 oO gt N o A G lt N Q Ss Q O Q oO N Code Meaning Comments 48 00 Initiator Detected Error An Initiator Detected Error message has been received and the message received previous phase the phase in which ATN was asserted was invalid 49 00 Invalid message error This is set for a number of reasons related to the host and drive not recognizing each others messages 4A 00 Command phase error A command could not be executed because too many parity errors occurred during the Command phase 4B 00 Data phase error A command could not be executed because too many parity errors occurred during the Data In and Data Out phases 4E00 Overlapped commands A host has selected the drive even though it already has a attempted command outstanding Dh VOLUME OVERFLOW Code Meaning 00 02 End of Tape detected Comments A WRITE or WRITE FILEMARKS command has encountered EOT or the physical end of tape The EOM flag will be set Note 60 Sense Keys and Codes When the Sense Ad
63. on separate buses the effective available bandwidth can be doubled Use one HBA for disks and put the tape drive on a separate bus Host CPU Speed Recommendation Faster hosts can typically transfer data quicker Use as fast a processor as possible for the backup system Network Transfer Time If backup involves transferring data over the network network performance is offen a major bottleneck Even on the fast systems at present the maximum effective EtherNet bandwidth is only about 1 MB s and not all this bandwidth is usable However the introduction of 100 MB s networks will make this factor of much less significance Recommendation E For large datasets use a tape drive attached directly to the server in order to maximize throughput E Minimize the number of clients that require backup Factors Affecting Performance 33 SIUDUIO I3d nn Q Tay 2 Ke 7 ES o 2 3 a Performance Factor Application Throughput Detail Some applications move data much quicker than others Select your application with care Write Commands Do not interleave write commands with other commands such as READ POSITION and LOG SENSE Do not for example attempt to read the TapeAlert log page during a long write Drive Related Factors Performance Factor Drive s SCSI Performance Recommendation Detail In order to minimize SCSI bus loading the drive must execute its SCSI phases q
64. ontrol Ultrium drives have built in hardware data compression Backup applications should incorporate features to E Control the compression capability of the drive E Report the actual compression ratio achieved during backup operations The typical compression ratio achieved during backup operations on PC and UNIX networks is 2 1 but this can vary widely depending on the actual data being compressed For more information see Controlling Data Compression on page 39 Non Immediate Commands Performance can be improved by only using immediate Write Filemarks commands Note Using immediate mode with other commands does not improve performance and can cause problems when writing a driver The SCSI specification requires that if a command is issued with the IMMEDIATE bit set to O the drive must flush its data buffer before it carries out the operation This takes time Managing the Use of Tapes The Ultrium format enables applications to monitor the performance of tapes closely to indicate when tape heads need cleaning and when a tape should be discarded See Use of Tapes on page 23 for more information Designing Backup Applications The Tape Log The LTO Cartridge Memory holds the System Area for each Ultrium tape that contains data about the tape s history This Tape log can be used to calculate the error rate history for the tape which in turn helps to determine the probability of a successful backup
65. pe failure On read Additional Sense VONNh Decompression exception On read otherwise Read from tape failure 4h HW ERROR Action Console Message This indicates that the current I O operation has failed due to a hardware failure The FRU code in the sense data should indicate which part of the hardware is bad The drive should not be used again until corrective action has been taken Specific recovery depends on the operating system and application For additional sense codes see 4h HW ERROR on page 53 Whatever the additional sense code the software should log the error terminate the I O and report the appropriate error to the calling application Whether the drive requires any further corrective action before it can be used again such as a reset or manual intervention depends on the additional sense code Tape drive hardware failure 5h ILLEGAL REQUEST Action Console Message The last command sent to the drive or the data sent because of the command could not be accepted by the drive because it violated conditions imposed by the drive For additional sense codes see 5h ILLEGAL REQUEST on page 54 Also see the Illegal Command Illegal Field Fixed Bit Reservation and Parameter List pre execution checks The software can retry the I O or else it can terminate the O and report an error to the calling application particularly if the I O has been retried a number of times and cont
66. persists call the tape drive fails its internal power on drive supplier s helpline self tests It is not cleared internally until the drive is powered off 32 Interface W The tape drive has a problem with the The drive has identified an application client interface interface fault 1 Check the cables and cable connections 2 Restart the operation 33 Eject media C The operation has failed Error recovery action 1 Eject the tape or magazine 2 Insert the tape or magazine again 3 Restart the operation 34 Download fail W The firmware download has failed Firmware download failed because you have tried to use the incorrect firmware for this tape drive Obtain the correct firmware and try again 35 Drive humidity W Environmental conditions inside the tape The drive s humidity limits have drive are outside the specified humidity been exceeded range 36 Drive W Environmental conditions inside the tape The drive is experiencing a temperature drive are outside the specified cooling problem temperature range Exception Handling 69 No Flag Type Recommended Host Message Cause 37 Drive voltage W The voltage supply to the tape drive is Drive voltage limits have been outside the specified range exceeded 38 Predictive C A hardware failure of the drive is Failure of the drive s hardware failure predicted Call the tape drive supplier s is predicted helpline 39 Diagnostics W The tape drive may have a hardware The drive
67. r following the text Ultrium A 1 indicates format generation 1 200 GB capacity at 2 1 compression and so on E Preferred method Use the SCSI Report Density Support command For a generation 1 product the following will be returned Primary Density Code 42h Vendor unique Density Code Assigning Organization LTO CVE Linear Tape Open Compliance and Verification Entity Density Name u28 Ultrium Generation 2 8 channel Configuration and Initialization 21 a te EEA a 2 c RS 9 o 53 sa Ss 2 Support for Additional LUN Enabling Additional LUN Support When enabled by an internally connected autochanger device an extra Logical Unit Number LUN will be available at the target s SCSI ID This allows the attached autochanger device to be addressed via the tape drive See Automation Control Interface ACI The normal tape drive LUN will always be LUNO and the automation LUN will always be LUN1 No other LUNs are available on the drive although HP is looking to provide new functionality throught the use of additional LUNs in future products Supporting Additional LUNs When working with a library vendor who is incorporating HP Ultrium drives in products software developers should liaise directly with the vendor about the functionality of the hardware available through the ACI 22 Configuration and Initialization Use of Tapes The HP Ultrium user documentation and Cartr
68. re may also be data associated with a set flag in the remainder of the flag byte which should also be read For each flag that is set log the associated error message Notify the user through the normal notification methods such as broadcast E Mail SNMP by displaying the error message suggested in the table Include the following Details to identify which drive the message refers to The software label of the tape cartridge when relevant The severity of the message Information Warning or Critical with Critical the most severe Where there are several flags set list the messages in ascending order of severity m x a d f 5 aS f 3 Ss a 4 Apply any error message filters in the software to the TapeAlert errors If several TapeAlert flags are set they should if possible be presented to the user as a single event For example the error messages could be displayed together in a single message box Exception Handling 73 5 Optionally automate the recommended recovery actions if there are multiple tape drives or autoloaders present For example the application could perform a cleaning cycle in response to flags 20 Clean Now and 21 Clean Periodic It could perform a tape copy for flags 4 Media performance degraded and 7 Media life expired and then retire the suspect tape cartridge This provides an opportunity for applications to add value to the TapeAlert capability of the drive
69. rmation will normally be supplied 1 2 4 Terminate the current I O and return the appropriate error The software should disable all further transactions to the drive and mark the drive as bad The software should tell the user that a serious fault has been detected with the drive and advise them to call their technical support Log the incident in the system log OC OB Media Auxiliary Memory write error An error has occurred while attempting to write to MAM The cartridge should not be used for further backups but should be able to be used for restoring data 1100 Unrecovered read error Action A read from tape has failed This is probably due to bad media but may be hardware related 1 2 4 Terminate the current O and return the appropriate error Send a console message that an unrecovered error on write has occurred Determine whether the error is deferred and report the last successful operation and the failed operation to the calling application Log the error and all recovery actions in the system log Recovery action is as follows 1 Use LOG SENSE to find the age and state of the tape and the drive Based on this ask the user to clean the drive or replace the tape If the fault is drive related ask the user to retry the operation after the drive has been cleaned If the fault is with the media prompt the user to back up the data to a new tape restart the application a
70. s Note An application must not fail a backup job as a result of TapeAlert information It should use the information as a preventative measure taking action to avoid failure or encouraging the user to take action It should also retain the log information to help in diagnosis if a job does eventually fail Responding to the Clean LED 74 Exception Handling Note We recommend that software applications use the TapeAlert log which should mean that potential tape or cleaning problems are flagged and corrected before the drive ever reaches the point of displaying the Clean LED If during normal operation the drive detects an excessive number of RAW retries the Clean LED is lit If this happens a user should follow this procedure 1 2 Clean the heads and try the operation again If the Clean LED is lit again repeat the operation with another tape cartridge If this clears the Clean LED it indicates that the original cartridge is at fault Copy the data from the cartridge onto a new one and discard the old cartridge The Clean LED is cleared by a cleaning cycle or by power cycling Providing Pass Through Mode It is important for Drivers and Logical Device Managers to provide a pass through mode that can return information or commands specific to a vendor s product The need for this is two fold E Systems must support a great variety of new devices m For the most part
71. software integration hp ultrium drives technical reference manual generation 2 SCSI and FC drives volume 2 software integration guide Part Number C7379 90900 Volume 2 Edition 2 February 2003 Notice The information contained in this document is subject to change without notice Hewlett Packard makes no warranty of any kind with regard to this material including but not limited to the implied warranties of merchantability and fitness for a particular purpose Hewlett Packard shall not be liable for errors contained herein or direct indirect special incidental or consequential damages in connection with the furnishing performance or use of this document Revision History This document contains proprietary information which is protected by copyright All rights reserved No part of this document may be photocopied reproduced or translated to another language without the prior written consent of Hewlett Packard Copyright 2003 by Hewlett Packard Limited Version Date Changes Edition 1 All HTML Generation 1 version Edition 2 Feb 2003 All PDF Generation 2 SCSI and FC drive version This document is frequently revised and updated To find out if there is a later version please ask your HP OEM Representative The Purpose of this Manual This is one of five volumes that document HP Ultrium drives This volume provides background information for driver and application developers The followi
72. t allow host software to suggest appropriate corrective action to the user For example if Flag 20 Clean Now is set the software should advise the user to clean the tape heads See the Chapter 4 of Volume 3 The SCSI Interface of the HP Ultrium Technical Reference Guide for details of the TapeAlert log page Flags The following table lists the flags that could potentially be supported in tape drives Of these flags 3 4 5 6 20 22 and 31 and mandatory for drives that support cleaning cartridges as do Ultrium drives The flags are grouped into the following sections E Flags 1 to 19 For tape drive write read management E Flags 20 to 25 For cleaning management E Flags 26 to 39 For tape drive hardware errors E Flags 40 to 49 For tape autoloader errors For each flag the message that the host software should display to the user is given together with the cause of the flag being set The Type column classifies the flags by seriousness into the following three groups Information A suggestion to the user m x a d f 5 aS f 3 Ss a W Warning The user is advised to take action Performance or data may be at risk otherwise C Critical Take action immediately Exception Handling 63 No Flag Type Recommended Host Message Flags for Tape Drive Write Read Management Cause 1 Read warning WwW The tape drive is having problems reading data No data has been lost bu
73. t combination to report that particular error For example a drive will report that it is not ready when there is no cartridge present by setting the sense key to NOT READY with additional sense of 0402h LUN not ready initializing command required No other combination of sense key and additional sense may be used to report that particular condition Sense Keys and Codes 45 Oh NO SENSE The following action applies to most additional sense codes in this group Action For all Additional Sense codes except 82 82h the action of the software depends on the current I O and what the operating system has been expecting Recovery depends on the operating system As a minimum the software should pass an error to the calling application indicating the positional mark that has been encountered The I O can be retried if desired Code Meaning 0000 No additional sense info Comments The drive has no additional sense information for the host The flags in the sense data indicate the reason for failure Action see above 0001 Filemark detected This indicates one of the following E A READ or SPACE command was terminated early because a filemark was encountered E Unsolicited Positional Sense has been set to indicate at a filemark The Mark bit in the sense data will always be set Action see above 00 02 End of tape detected A command completed early because End of Tape or the physica
74. t there has been a reduction in the capacity of the tape The drive is having severe trouble reading 2 Write warning W The tape drive is having problems writing data No data has been lost but there has been a reduction in the capacity of the tape The drive is having severe trouble writing 3 Hard error WwW The operation has stopped because an error has occurred while reading or writing data which the drive cannot correct This flag is set for any unrecoverable read write positioning error and is cleared internally when the tape is ejected The flag is set as an explanation of the error in conjunction with one of the recovery action flags 4 5 or 6 4 Media C Your data is at risk Media performance is severely 1 Copy any data you require from this degraded or the pecan nip ipe longer be written or read This flag is set for an t use this t 9 y Donel dpe Mie tape again unrecoverable read write Restart the operation with a different positioning error caused by tape faulty media It is cleared internally when the media is ejected 5 Read failure C The tape is damaged or the drive is The drive can no longer read faulty Call the tape supplier s helpline data from the tape The flag is set for any unrecoverable read error where the diagnosis is uncertain and could either be a faulty tape or faulty drive hardware It is cleared internally when the tape is ejected 64 E
75. tains techniques and information to help you design software applications so that they use the tape drive s potential as efficiently as possible Techniques of optimizing performance Detecting the drive s speed Ensuring the recommended minimum transfer sizes Using Cartridge Memory instead of tape headers Tuning performance using the Performance Log page Time out values to help you tune timings in backup applications E Log pages This page gives recommendations on the use of information from the log pages E Factors affecting performance This page examines factors relating separately to the host the drive and the format and how they can affect performance Ways of Optimizing Performance HP s Ultrium drives are high performance products The drive s native speed is 30 MB s With a typical compression ratio of 2 1 this doubles to 60 MB s Application software may require significant enhancement in order to capitalize on this speed There are a number of areas to look at and these are discussed below Further details can also be found in the How to optimize the performance of hp ultrium tape drives white paper Factors Affecting Performance 29 Detecting the Drive s Speed Applications should not key off Inquiry strings in order to tell the difference between different speed drives It is better to use the Performance Log page see under the LOG SENSE command in Chapter 4 Commands of the
76. tape has been successfully loaded and tape motion has ceased The remaining capacity value is the amount of tape remaining calculated from the current position As a result unless the tape is positioned at EOD the calculated value ignores the fact that there may be more data written further up the tape m Remaining capacity values are only valid after the successful completion of the following commands in non immediate mode LOAD LOCATE MODE SELECT READ REWIND SPACE VERIFY WRITE WRITE FILEMARKS The values after any subsequent command cannot be relied on unless the command is a sense type that does not cause any tape motion Calculations of capacity are inaccurate in two ways E Random errors can be caused by tolerances in tape length hub diameter and so on These are described below E Systematic errors are caused by ignoring system areas and so on They ensure that the calculated capacity is actually available to the user It is usually possible to write considerably more data than the calculated capacity Tape Length Tolerance Random Errors Tape length is subject to significant tolerances in its calculation These derive from tape thickness capstan diameter supply reel hub diameter and supply reel velocity measurements Tolerances vary from BOM to EOM due to the increasing effect of the supply reel hub diameter tolerance The existence of these tolerances can cause apparent anomalies in returned values E Successive
77. the correct format tape 2 Discard the tape or return the tape to your supplier bulk erased or is of an incorrect format 55 Loading failure C The operation has failed because the media cannot be loaded and threaded 1 Remove the cartridge inspect it as specified in the product manual and retry the operation 2 If the problem persists call the tape drive supplier s help line The drive is unable to load the cassette and thread the tape 56 Unrecoverable load failure C The operation has failed because the tape cannot be unloaded 1 Do not attempt to extract the tape cartridge 2 Call the tape driver supplier s help line The drive is unable to unload the tape 57 Automation interface failure C The tape drive has a problem with the automation interface 1 Check the power to the automation system 2 Check the cables and cable connections 3 Call the supplier s helpline if the problem persists The drive has identified a fault in the automation interface 58 Firmware failure 4 The tape drive has reset itself due to a detected firmware fault If the problem persists call the supplier s helpline There is a firmware bug Flags 59 64 are not currently used Exception Handling 71 m x a d 5 ae 2 Ss Ss a Note that often messages will not appear in isolation For example message Olh The tape drive is having
78. the data again if an error is found Small Computer System Interface a standard command specification and command set that enables computers and peripherals to communicate with each other HP s Ultrium drives adhere to the SCSI specifications see Chapter 1 Interface Implementation in Volume 3 The SCSI Interface of the HP Ultrium Technical Reference Manual and support all features required by those standard Single Ended and Low Voltage Differential SCSI These terms define how the signals are transmitted along the cable With single ended SE SCSI each signal travels over a single wire and each signal s value is determined by comparing the signal to a paired ground wire Signal quality tends to decrease over longer cable lengths or at increased signal speed With low voltage differential IVD signaling signals travel along two wires and the difference in voltage between the wire pairs determines the signal value This enables faster data rates and longer cabling with less susceptibility to noise than SE signaling and reduced power consumption Narrow and Wide Fast Ultra and Ultra2 SCSI Narrow SCSI devices can transfer data one byte at a time and are sometimes called 8 bit SCSI devices They can conform to either the SCSI 2 or SCSI 3 protocols They have a 50 pin connection to the SCSI bus Wide SCSI devices can transfer two bytes of data simultaneously 16 bit SCSI They usually have a single 68 pin
79. ther state at which point either execute the command or terminate the I O 3h MEDIUM ERROR Code Meaning Comments 0002 End of tape detected A READ SPACE WRITE or WRITE FILEMARKS command found EOT unexpectedly This typically occurs when a drive cannot locate the target object on tape because the block count is too great The EOM flag will be set Action 1 Recovery action depends on the initiating action As a minimum tell the calling application that physical EOP M has been encountered Also display this information as a console message n e gt N ko A G lt N Q Ss Q O Q oO N 2 Send any residue information to the calling application 0004 Beginning of tape detected A SPACE command terminated early because it reached BOT This typically occurs when a drive cannot locate the target object on tape because the block count is too great The EOM flag will be set Action 1 Recovery action depends on the initiating action As a minimum tell the calling application that physical BOP has been encountered Also display this information as a console message 2 Send any residue information to the calling application Sense Keys and Codes 49 Code Meaning oc 00 Write error Action Comments The drive has failed to write data or filemarks to tape This is probably due to bad media but may be hardware related Separate additional sense is reported in the case of an Erase failure Residue info
80. uickly The phases are as follows Selection Message out identification Receipt of the command Disconnection Mid command bus free time Arbitration and reselection Message in identification The host must always ensure the following E Disconnects are enabled m Synchronous negotiation is enabled and established between the drive and the HBA E The drive is in buffered mode m When reading and verifying always use the same block size as that in which the tape is written otherwise performance will be very seriously affected Transfer Mode Recommendation The transfer mode can be fixed or variable selectable through the MODE SELECT command Fixed Mode The transfer size is equal to the block size multiplied by the number of records blocks in the transfer A good transfer size to aim at is 64 kilobytes For an application that uses 512 byte records each fixed mode transfer should transfer 128 records Variable Mode Only one block is transferred at a time The size of the block determines the size of the transfer 34 Factors Affecting Performance Performance Factor Recommendation Detail In variable block mode the application should use 64 kilobyte blocks Records Block Size Recommendation The size of the transfer impacts the performance rather than the size of the record blocks in the transfer As above aim to use 64 kilobyte transfers Transfer Size Recommendation
81. urned in a multi host environment 56 Sense Keys and Codes Code Meaning Comments Action When operating the drive in this type of environment the following actions should occur 1 The calling application receiving this code should issue a MODE SENSE command requesting the drive to return all parameters 2 The application should check those parameters over which it has configuration control to ensure that the current configuration of the drive does not conflict with what the application expects 3 If it finds discrepancies the application can either reconfigure the drive to the original values or halt and report an error 4 Ifan error is reported a console message must be displayed and information logged to the system log 3F01 Microcode downloaded The firmware in the drive has just been changed by a WRITE BUFFER command 3F 11 Media Auxiliary Memory MAM is accessible but the cartridge is in the Hold position accessible 55 06 Media Auxiliary Memory full There is insufficient space in the Host Attribute area in MAM to fit the y attribute that need to be written 2 A Action Check MAM attribute 0004h MAM Space Remaining to identify 3 how much space remains in MAM 2 em 5D FF False informational A TapeAlert Mode page has been sent with the Test field set to 1 S exception condition and the DExcpt field to O causing the drive to generate a false 2 generated informational exception condition a false device f
82. user of existing or impending conditions and can for example advise the user to change the tape glossary 81 82 glossary A ACI 22 additional sense codes 45 ANSI 3 79 autochangers 22 B backup applications 13 block mode fixed length 13 variable length 14 block size maximum 30 BOM 79 BOT 49 buffer maximizing use 13 size at EW EOM 40 buffered mode 79 C capacity of tape 23 remaining 24 cartridge memory 23 37 using instead of headers 30 cartridges see tapes Clean LED 74 cleaning 46 74 cleaning cartridge 48 cleaning tape heads 15 commands non immediate 14 compression 79 controlling 39 ratio 79 configuration 19 D data compression 79 controlling 13 14 39 effect on performance 35 data transfer size 13 decompression 79 design goals 17 Diagnostic logs 16 diagnostics failure 54 displaying drive information 16 documents related 3 drive checking integrity 17 detecting speed 30 displaying information 16 initialization 19 monitoring condition 63 drive technology family 21 drivers 19 E ECMA 79 eject failed 53 54 EOD 79 detected 58 not found 59 EOM 46 49 79 EOT 60 erase failure 54 errors hard 80 parity 59 read 50 write 50 52 escalation procedure 61 Index 83 84 Index EW EOM 79 exception handling 61 F faults predicting 16 62 filemarks 79 count 80 detected 46 use of 36 firmware revision 16 fixed length block mode 13 flags TapeAlert 63 for
83. xception Handling No 6 Flag Write failure Type C Recommended Host Message The tape is from a faulty batch or the tape drive is faulty 1 Use a good tape to test the drive 2 If the problem persists call the tape drive supplier s helpline Cause The drive can no longer write data to the tape The flag is set for any unrecoverable write positioning error where the diagnosis is uncertain and could either be a faulty tape or faulty drive hardware It is cleared internally when the tape is ejected 7 Media life W The tape cartridge has reached the end The media has exceeded its of its calculated useful life specified life 1 Copy any data you need to another tape 2 Discard the old tape 8 Notdata grade W The tape cartridge is not data grade A DDS drive has not been able pol rilevante Any data you write to the tape is at risk saa T a a that Ultrium drives Replace the cartridge with a data grade NAICS cata grade tape tape 9 Write protect C You are trying to write to a write A write command was protected cartridge attempted to a write protected Remove the write protection or use tape another tape 10 No removal You cannot eject the cartridge because A manual or software unload the tape drive is in use was attempted when Prevent Wait until the operation is complete Medium Removal was in force before ejecting the cartridge 11 Cleaning media The tape in the drive is a cleaning

Software - Oracle Documentation

Contents

Download Pdf Manuals

Related Search

Related Contents