Hello list, I'm implementing disk-to-disk-to-tape backups with bacula 3.0.2, using copy jobs to duplicate the disk-to-disk jobs to tape for off-site storage. I'm running into a problem when I manually run the copy job in that it only copies the *last* of the set of job id's to tape.
I'm using this suggestion [1] from the bacula-devel list to copy the disk-to-disk jobs into my tape pools, which cycle every four weeks and are written on mondays, wednesdays and fridays. I wrote a postgres stored procedure which locates the correct job IDs for a host to use in the selection pattern[2]. The tape pools are configured as in [3], and the copy job as in [4]. I've not got to the point of running these jobs from schedules yet, as I am still in the testing phase of the disk-to-tape work. I am running the copy job with: run pool=W01-copy <hostname>-tape I can see that my stored procedure is returning the correct job id's. Here is an example run's log output: 29-Oct 09:46 bksrv0-dir JobId 105: The following 4 JobIds were chosen to be copied: 38,53,82,101 29-Oct 09:46 bksrv0-dir JobId 105: Job not run. 29-Oct 09:46 bksrv0-dir JobId 105: Error: Could not start migration job. 29-Oct 09:46 bksrv0-dir JobId 105: Job not run. 29-Oct 09:46 bksrv0-dir JobId 105: Error: Could not start migration job. 29-Oct 09:46 bksrv0-dir JobId 105: Job not run. 29-Oct 09:46 bksrv0-dir JobId 105: Error: Could not start migration job. 29-Oct 09:46 bksrv0-dir JobId 105: Copying using JobId=101 Job=<hostname>.2009-10-28_21.00.00_41 29-Oct 09:46 bksrv0-dir JobId 105: Unable to get Job Volume Parameters. ERR=sql_get.c:442 No volumes found for JobId=101 29-Oct 09:46 bksrv0-dir JobId 105: Previous Job has no data to copy. <snip> The last error is not a problem; the last incremental for the host in question had no data to back up. The problem is the three "Job not run." errors above that. I have looked through the source[5] and here's where I am seeing what looks like a bug. On line 867 all but the last of the jobs is passed on to start_migration_job(). On line 920-921, inside this subroutine, a new run command line is assembled, and on line 24 it is run. However, what is *not* being passed on is my initial "pool=W01-copy", which causes the job to fail. I have tested this in bconsole too; running "run pool=W01-copy jobid=53 job=<hostname>-tape" ran a copy job fine, whereas running "run job=<hostname>-tape jobid=53" failed with the same "Job not run." error. In my schedules (that are not live yet) the pool will be specified. I can't figure out whether this means that that pool will be passed on to the jobs started by start_migration_job() correctly, despite having spent most of the morning looking at the code. I also can't offer a patch to fix the problem; I wouldn't know where to start :-) If I can provide any more useful information to help fix the problem, please just ask! Many Thanks, --Alex [1] http://www.mail-archive.com/[email protected]/msg04599.html [2] CREATE OR REPLACE FUNCTION find_backup_jobs(host varchar) RETURNS SETOF integer AS $$ DECLARE last_full_time integer; last_diff_time integer; BEGIN SELECT max(jobtdate) INTO last_full_time FROM job WHERE name = host AND level = 'F' AND type = 'B'; SELECT max(jobtdate) INTO last_diff_time FROM job WHERE name = host AND level = 'D' AND type = 'B' AND jobtdate > last_full_time; IF last_diff_time IS NOT NULL THEN -- differentials exist so this is a monthly backup RETURN QUERY SELECT jobid FROM job WHERE name = host AND type = 'B' AND ( level = 'F' and jobtdate = last_full_time OR level = 'D' and jobtdate = last_diff_time OR level = 'I' and jobtdate > last_diff_time); ELSE -- no differentials here so grab full and all incrs RETURN QUERY SELECT jobid FROM job WHERE name = host AND type = 'B' AND ( level = 'F' and jobtdate = last_full_time OR level = 'I' and jobtdate > last_full_time); END IF; END; $$ LANGUAGE plpgsql; [3] (an example) Pool { Name = W01 Pool Type = Backup Auto Prune = Yes Recycle = Yes Storage = TS3100 Volume Retention = 27 days Recycle Oldest Volume = yes } Pool { Name = W01-copy Pool Type = Backup Next Pool = W01 } [4] (this comes from a perl script that generates bacula configuration from a YAML input file) Job { Name = "$hostname" Client = "$hostname" JobDefs = "default-job" Storage = "$hostname" Pool = "$hostname" Schedule = "$schedule" FileSet = "$fileset" } Job { Name = "$hostname-tape" Client = "None" JobDefs = "default-tape" Pool = "None" # overridden by schedule Storage = "$hostname" # Schedule = "copy-to-tape" Selection Pattern = "select jobid from find_backup_jobs('$hostname') as jobid;" } [5]: http://bacula.git.sourceforge.net/git/gitweb.cgi?p=bacula/bacula;a=blob;f=bacula/src/dird/migrate.c;h=2ffadb76fc04d502d81d4f33dfee6d96960b9ed8;hb=HEAD ------------------------------------------------------------------------------ Come build with us! The BlackBerry(R) Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9 - 12, 2009. Register now! http://p.sf.net/sfu/devconference _______________________________________________ Bacula-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/bacula-devel
