Andrzej Zawadzki schrieb:
My bacula just finished full backup of last server:7488 Full 58,075 1.525 G OK 21-lut-08 07:36 NSwiatrak now time is: Thu Feb 21 08:57:08 CET 2008 and bacula is waiting... Running Jobs: JobId Level Name Status ====================================================================== 7489 Tape-Eject.2008-02-19_04.00.55 is waiting for higher priority jobs to finish 7490 Full BackupCatalog.2008-02-19_04.30.56 is waiting execution 7500 Full BackupCatalog.2008-02-20_04.30.06 has been canceled 7501 Increme NSspichlerz.2008-02-21_00.05.11 is waiting for a mount request
Ah yes, that's a cute one. I have run into something similar already. It does seem a bit counterintuitive at first, to put it mildly. :-) What has happened here is that by the time your 2008-02-19 job had finished, the job scheduled for 2008-02-21 was already due. As your jobs "Tape-Eject" and "BackupCatalog" have lower priority than your actual backup job "NSspichlerz" (as recommended in the Bacula manual) job 7501 "skipped the queue" and was started before the earlier jobs 7489 and 7490. So job 7501 is now waiting for its tape, but in vain, because the previous tape has not been ejected. But job 7489 which should have done that is in turn waiting for job 7501 because that one had higher priority, IOW "it should go first". Deadlock. And the catalog job 7490 is caught in between. Btw, I think it is a mistake to put the BackupCatalog job after the Tape-Eject job. How can it backup the catalog if the tape has already been ejected? How to resolve the deadlock? Two possibilities: a) If you want to complete the 2008-02-19 backup correctly: Cancel job 7489 and any jobs after 7501 that might already be in a "waiting" state. Then cancel job 7501. Job 7490 will be the next to run, and back up your catalog, so the backup of 2008-02-19 will be complete. Then unmount and eject the tape manually, mount the next tape, and let the cycle resume. b) If you prefer to have the 2008-02-21 job run now: Cancel first job 7490 (so that it can't interfere), then 7489, unmount and eject the tape manually, cancel any jobs for dates after 2008-02-21 that may already be in "waiting" state, and then mount the next tape. Job 7501 will resume, and afterwards everything will hopefully proceed according to your schedule. How to avoid that in the future? I have put that question to the list once already. The only answer was to run jobs that might require operator intervention (read: any backup job) only at times when an operator is present (read: not on weekends or holidays). IOW: Bacula doesn't like waiting. :-) HTH T. -- Tilman Schmidt Phoenix Software GmbH www.phoenixsoftware.de 53227 Bonn, Germany Amtsgericht Bonn HRB 2934
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
