Sybase backup failure

[ES UNIX Team]

All,
I have been dealing with this for several weeks now. We have replaced the tapes, tape drives and SCSI cables. The latest SDLT220 patches are installed. I am running a cluster (Sybase only runs on the active member) and have failed it over which should rule out SCSI cards unless we assume the SCSI card is bad on both servers. My opinion is that there is a problem with Sybase or the OS, but the DBA Team insists it must be hardware. Anyway, the Sybase error logs report "error number 5 (I/O error)" followed by "Msg 8009, Level 16, State 1". The first error is a general OS error and the later is a general Sybase error. The servers are running HP-UX 11.0 with Sybase 12.5 using the Sybase internal backup service. When the backups fail they usually are able to write some data to tape before failing (it vaires, anywhere fom 1 GB to 200 GB). The tapes are always rewound so it isn't a space issue. The syslog file never reports any errors. Every day after a failure I am able to re-insert the same tape and write to it with basic OS tools (tar, dd, etc.). Therefore I believe this is a Sybase issue, but I wouldn't know what to look at and I can't convince the database team that it is. They insist it must be hardware. Have I missed anything? Do you have any thoughts for anything I can do more with hardware/OS? Have you seen this with Sybase before and if so can you point to a specific issue that can be addressed? Thanks in advance for your help.

[RAC_1]

Does this sybase internal backup utility writes to some temporary area and then to tape?

Do you see any messages fot tape drive in dmesg/syslog/stm??

Can you backup from OS prompt to tape without problems?

Anil

[Michael Schulte zur Sur]

Hi,

try with
dd if=/dev/... of=/dev/null
the devices in question. Hardware is either disk or tape. Are you using raw devices?

greetings,

Michael

[RAC_1]

Just another thought.

Do a mediainit /dev/rmt/xxx before you start taking backup.

Anil