Menu
▾
▴
[Bacula-bugs] [bacula 0000927]: Multi drive changer keeps switching which drive a job runs on and blocks because the tape is in the other drive.
|
From: <bac...@li...> - 2007-08-21 17:36:11
|
A NOTE has been added to this issue. ====================================================================== http://bugs.bacula.org/view.php?id=927 ====================================================================== Reported By: drescherjm Assigned To: ====================================================================== Project: bacula Issue ID: 927 Category: Storage Daemon Reproducibility: sometimes Severity: minor Priority: normal Status: feedback ====================================================================== Date Submitted: 08-21-2007 11:45 EDT Last Modified: 08-21-2007 13:36 EDT ====================================================================== Summary: Multi drive changer keeps switching which drive a job runs on and blocks because the tape is in the other drive. Description: I upgraded to bacula-2.2.0 last week and every night the storage daemon gets stuck waiting for a mount request on an autochanger when the requested tape is already in a drive of the changer but not the drive that the storage daemon wants. This has happened 3 nights in a row and it did not occur with bacula-2.1.X (not sure of the exact version of the sd at the time of the upgrade). ====================================================================== ---------------------------------------------------------------------- drescherjm - 08-21-07 11:49 ---------------------------------------------------------------------- BTW, The pool LTO2-Study-Backup is not the pool of this tape A00015 as it should be UserBackup-LTO2. ---------------------------------------------------------------------- kern - 08-21-07 13:23 ---------------------------------------------------------------------- This is a bit surprising because it is exactly the kind of situation that 2.2.0 *should* handle much better than 2.0.x Your note about the pool difference is probably the key to the problem. Could you restart all the daemons, and run the SD with debug level set to 51 and the output directed to a file? When it blocks again, please make a copy of the file and attach it to this bug report. It should allow me to find out why you are having this problem. It would also be helpful to have the Job reports of all the jobs that did run during the debug period. ---------------------------------------------------------------------- drescherjm - 08-21-07 13:36 ---------------------------------------------------------------------- Thanks Kern. I will do that after the waiting jobs finish. I issued mount on the first drive and it appears bacula is continuing with the backups: 3001 OK mount. Device="LTO2-0" (/dev/nst0) 21-Aug 13:31 dev6-sd: 3301 Issuing autochanger "loaded? drive 0" command. 21-Aug 13:31 dev6-sd: 3302 Autochanger "loaded? drive 0", result: nothing loaded. 21-Aug 13:31 dev6-sd: 3301 Issuing autochanger "loaded? drive 0" command. 21-Aug 13:31 dev6-sd: 3302 Autochanger "loaded? drive 0", result: nothing loaded. 21-Aug 13:31 dev6-sd: 3301 Issuing autochanger "loaded? drive 0" command. 21-Aug 13:31 dev6-sd: 3302 Autochanger "loaded? drive 0", result: nothing loaded. 21-Aug 13:31 dev6-sd: 3307 Issuing autochanger "unload slot 17, drive 1" command. Issue History Date Modified Username Field Change ====================================================================== 08-21-07 11:45 drescherjm New Issue 08-21-07 11:49 drescherjm Note Added: 0002690 08-21-07 11:57 drescherjm Issue Monitored: drescherjm 08-21-07 13:23 kern Note Added: 0002691 08-21-07 13:23 kern Status new => feedback 08-21-07 13:36 drescherjm Note Added: 0002692 ====================================================================== |
