Re: [Bacula-users] Multiple Restore jobs
Hi Bill, You are correct,, I can see the MaximumConcurrentJobs limitation, see the "status dir" output from bconsole below: So how to we enhance this limit for the Restore jobs Connecting to Director bacula-server:9101 1000 OK: 103 bacula-server-dir Version: 9.4.4 (28 May 2019) Enter a period to cancel a command. status dir bacula-server-dir Version: 9.4.4 (28 May 2019) x86_64-redhat-linux-gnu-bacula redhat Daemon started 10-Apr-23 16:35, conf reloaded 14-Apr-2023 14:32:33 Jobs: run=3557, running=7 mode=0,0 Heap: heap=6,049,792 smbytes=88,029,733 max_bytes=116,303,584 bufs=498,248 max_bufs=501,721 Res: njobs=947 nclients=533 nstores=21 npools=83 ncats=1 nfsets=91 nscheds=34 Scheduled Jobs: Level Type Pri Scheduled Job Name Volume === Running Jobs: Console connected at 10-Apr-23 16:41 Console connected at 12-Apr-23 11:39 JobId Type Level Files Bytes Name Status == 411391 Back Incr 0 0 varunama is waiting for Client to connect to Storage userStorageF 411399 Back Incr 14370.0 M mmadhavi is running 411410 Back Incr 0 0 a10a is waiting for Client to connect to Storage userStorageE 411419 Back Incr 0 0 naha is waiting for Client to connect to Storage userStorageB 411421 Back Incr 0 0 satyprak is running 411455 Rest Rest 21,7016.108 G RestoreFiles is running 411456 Rest Rest 0 0 RestoreFiles is waiting on max Job jobs Terminated Jobs: JobId Level FilesBytes Status FinishedName You have messages. Thanks Yateen -Original Message- From: Bill Arlofski Sent: Thursday, April 13, 2023 7:43 PM To: Yateen Shaligram Bhagat (Nokia) ; bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Multiple Restore jobs On 4/13/23 03:51, Yateen Shaligram Bhagat (Nokia) wrote: > Thanks Bill, that helps. > > There is a reason why we ned to run multiple restore jobs concurrently. > > We are migrating our computing env (few hundred hosts) to a different > flavour of Linux. > Prior to migration we need to do data backup and then restore it back. > > If we go sequentially using one single RestoreFiles job, it is going to be a > lengthy exercise. > > Anyway I will try defining and running multiple restore jobs. Yateen, Again, you only need *one* Restore Job defined, even if you are going to run 1,000 restores at the same time. Show us the problem you are experiencing so we can direct you to the right solution. I am guessing you are more than likely running into some "MaximumConcurrentJobs" bottleneck somewhere. I mean, show us a Director that is running (or trying to run) multiple restore jobs at the same time: * status dir Best regards, Bill -- Bill Arlofski waa@pro tonmail.com ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Multiple Restore jobs
On 4/14/23 03:42, Yateen Shaligram Bhagat (Nokia) wrote: Hi Bill, You are correct,, I can see the MaximumConcurrentJobs limitation, see the "status dir" output from bconsole below: So how to we enhance this limit for the Restore jobs Hello Yateen, I think you just need to add `MaximumConcurrentJobs = x` to your restore job called `RestoreFiles` Set `x` to any number higher than 1, like 10, 20, whatever, and you should be OK to do multiple concurrent restore jobs. Hope this helps, Bill -- Bill Arlofski w...@protonmail.com signature.asc Description: OpenPGP digital signature ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Deleting Obsolete Volumes
The problem is that the catalog doesn't record which jobs are on each volume. It uses the jobmedia table to record which files of a job are on each volume (in groups between firstindex and lastindex). As a result, if a job has no files then it has no rows in the jobmedia table and hence is invisible to 'query 14'. Why are you worried about deleting these empty volumes? There should be no problem because they will never be used in a restore. __Martin > On Thu, 13 Apr 2023 23:17:50 +0100, Chris Wilkinson said: > > Yes that is true. In the other cases I found, these were also zero file > incr/diff jobs. > > However it appears that the job wrote a Volume as it appears in the list > media result and there is such a Volume on disk but only 667 bytes so only > the label I presume. > > I had expected that 'query 14' would recognise that a Volume has an > associated job even if that job wrote zero files to the volume but it > doesn't. > > The script needs some additional condition for deletion to avoid deleting > these zero file volumes. No idea what that might be right now. > > Chris- > > On Thu, 13 Apr 2023, 22:55 Bill Arlofski, wrote: > > > Hello Chris, > > > > Your jobid 2,371 wrote 0 files, 0 bytes, and the job's Summary shows for > > `Volume name(s):` No volumes. > > > > So, unless there are other jobs on this volume, jobid 2,371 did not > > actually write to it. > > > > > > Hope this helps, > > Bill > > > > -- > > Bill Arlofski > > w...@protonmail.com > > > > -Chris- > ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Multiple Restore jobs
Bill, Thanks , will try that ! -Yateen -Original Message- From: Bill Arlofski Sent: Friday, April 14, 2023 8:07 PM To: Yateen Shaligram Bhagat (Nokia) ; bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Multiple Restore jobs On 4/14/23 03:42, Yateen Shaligram Bhagat (Nokia) wrote: > Hi Bill, > > You are correct,, I can see the MaximumConcurrentJobs limitation, see the > "status dir" output from bconsole below: > So how to we enhance this limit for the Restore jobs Hello Yateen, I think you just need to add `MaximumConcurrentJobs = x` to your restore job called `RestoreFiles` Set `x` to any number higher than 1, like 10, 20, whatever, and you should be OK to do multiple concurrent restore jobs. Hope this helps, Bill -- Bill Arlofski w...@protonmail.com ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users