Hello, Bacula does not treat write errors as fatal -- it is an error, but Bacula simply ends the Volume and continues. If enabled, it will attempt to re-read the last good block written, and if it succeeds, it is a good indication that the Volume is perfectly good, though it may be truncated because of the write error.
So, what you are seeing below is perfectly normal. The only quark is the block number was not what it suspected. Since you didn't specify what version of Bacula you are using, I won't comment on it -- how Bacula keeps track what it accepts as the last valid block is version dependent. Regards, Kern On Saturday 28 March 2009 00:03:07 James Harper wrote: > I have a cabling problem with my scsi tape drives. I thought it was > fixed but obviously not. When doing my VirtualFull to tape last night, > Bacula says: > > 27-Mar 23:16 bitvs2-sd JobId 1235: Reposition from (file:block) > 10:1428374769 to 11:2752208962 > 27-Mar 23:19 bitvs2-sd JobId 1235: Reposition from (file:block) > 12:2457049971 to 12:2457049972 > 27-Mar 23:20 bitvs2-sd JobId 1235: Reposition from (file:block) > 12:3456985893 to 12:3456985894 > 27-Mar 23:21 bitvs2-sd JobId 1235: Error: block.c:568 Write error at > 53:2863 on device "TapeStorage" > (/dev/tape/by-path/pci-0000:01:08.0-scsi-0:0:6:0-nst-nst). > ERR=Input/output error. > 27-Mar 23:21 bitvs2-sd JobId 1235: Error: Re-read of last block OK, but > block numbers differ. Read block=421361 Want block=421362. > 27-Mar 23:21 bitvs2-sd JobId 1235: End of medium on Volume "BIT0011" > Bytes=51,874,099,200 Blocks=804,099 at 27-Mar-2009 23:21. > > And the kernel says: > > Mar 27 23:21:27 bitvs2 kernel: st1: Current: sense key: Aborted Command > Mar 27 23:21:27 bitvs2 kernel: Additional sense: Scsi parity error > > That's definitely not an end-of-medium. I would have expected that > Bacula would at least mark the tape itself as failed or something. > > What is supposed to happen when a scsi error occurs? > > Thanks > > James > > --------------------------------------------------------------------------- >--- _______________________________________________ > Bacula-devel mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/bacula-devel ------------------------------------------------------------------------------ _______________________________________________ Bacula-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/bacula-devel
