On Monday 18 September 2006 02:27, Francisco Reyes wrote: > Kern Sibbald writes: > > > The algorithm is very clearly described in the manual at: > > http:/www.bacula.org/dev-manual/Automatic_Volume_Recycling.html#SECTION000243000000000000000 > > In that url I see: > Try recycling any purged Volumes > > Does this meet the criteria? > recycle = 1 > Difference of current time minus last written is greater than volretention > column > > > > They do recycle based on the retention periods as far as I know. > > I have 484, which was only 358 yesterday, volumes that meet the criteria > above and I am just trying to determine if I am doing something wrong on how > I am researching this. Did I miss anything in that criteria? > > Example volume: > BaculaVolumeDaily0203 > lastwritten: 2006-08-03 > volstatus: Purged > recycle: 1 > volretention: 604800 > volretention/(3600*24): 7 days which matches my 7 day retention for daily > pool > (CURRENT_DATE-lastwritten::date)::int: 45 > > The volume is 45 days, is marked as recycle = 1, volstatus is Purged. > Shouldn't that volume have been used instead of having new volumes created > in that pool? > > I posted the queries that I used and would be nice if anyone else using > postgresql could give them a spin.. and see if others have volumes on > simmilar situation. > > > algorithm is not being followed, then I haven't seen clear proof of it. > > I honestly don't know if my criteria above is correct, but would like any > feedback to improve it and if I am doing something wrong hopefully fix it.
The above is what one could call "hearsay", not proof. See below ... > > > > few people from time to time say "you see it doesn't work", but without > > supplying all the necessary info. > > > What information would be helpfull? - Your version of Bacula - What database (MySQL, ...) you are using and its version. I see you mention that it is PostgreSQL. - A list of all the volumes in the pool. - An "llist pool=xxx" - An "llist volume=yyy" where you pick one representative volume, which should be the one with the oldest LastWritten date. - Your Job resource - Your Pool resource - The complete output of a job where recycling failed. > > > The problem is that the algorithm is rather > > complex and most users either don't understand or don't read all the little > > details. > > Understood.. > > I am trying to look into only one part of the whole algorithm.. from the > database perspective.. basically I am looking at volumes and their columns > .. and trying to see if any of those volumes should have gotten used instead > of new ones getting created. I don't exclude the possibility of a bug. It is just that after 10 or 20 false alerts (i.e. user errors), I don't get excited any more since I don't have the time to "debug" user problems and program Bacula. > > > The only thing I can say is that I have seen recycling work for 3 years now > > without fail, and if you read that chapter I am sure you can sit back like me > > and not worry for a long time. > > I am hopefull I am doing something wrong and that is just a matter of > keeping researching and some guidance, but so far I get the impression that > somehow many volumes that I believe (and I could be totally wrong) should be > re-used are not getting reused. > > > By the way, the database (SQLite 2.x) has remained stable at 170MB, > > My postgresql DB for bacula is 30GB with daily vacuum analyze and > autovacuum.. have lots of small files.. and have not even finished adding > all our servers to be managed by Bacula. > > > Full backups were running about 30GB, but after cleaning out old Photos and > > Music files, it dropped to about 10GB ... > > By the time I add all our servers we will be backing up likely close to > 1Terabyte. The very near future we need to backup 500G, so getting volumes > to be reused properly is a big issue for us. > > For reference the queries I am using, on postgresql, to research this are: > > -- Volumes I believe should have been re-used but arent > select volumename,mediatype,firstwritten::date,lastwritten::date, > labeldate::date,volstatus,recycle, volretention, > volretention/(3600*24) as Retention, (CURRENT_DATE-lastwritten::date)::int > as Age > from media where volstatus = 'Purged' and recycle = 1 and > volretention/(3600*24) < (CURRENT_DATE-lastwritten::date)::int > limit 10; > > > -- Count only > select count(*) > from media where volstatus = 'Purged' and recycle = 1 and > volretention/(3600*24) < (CURRENT_DATE-lastwritten::date)::int > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users > ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
