Re: [bareos-users] How to get parallel jobs working?
Hi Bruno, Am Freitag, 9. März 2018 19:28:34 UTC+1 schrieb Bruno Friedmann: > I guess you should invest so time to read and setup spooling. > I'm using that and I bundle easily 10 to 30 jobs on one media. That would (as far as I understand it) mean writing multiple jobs to the same Volume. As I have to be able to remove all backups for certain hosts at will, that's not possible here... But thanks anyways for the hint! Cheers, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
On mardi, 6 mars 2018 10.48:04 h CET Martin Emrich wrote: > Am Dienstag, 6. März 2018 01:23:45 UTC+1 schrieb Stefan Klatt: > > I have a few comments. > > > > - Do you have really monolithic config files? They are bad to read > > and old school :-). > > In reality the config files are split via the @ operator (one file for > pools, one for storage, one for job templates and schedules, and one for > each of ca. 100 clients in a subdirectory)... I just reduced it for better > readability and kept only one job for each pool. > > - I didn't know that the construct JobDefs -> JobDefs -> Job > > works. > > I had no other issues with that, but I'll try to flatten the hierarchy to > one JobDefs level. > > - For me your configuration is more complex than needed. E.g. > > Storage definitions at Pools, Jobs and JobDefs is too much. > > There is no Storage definition in the Jobs. But Indeed I could remove the > Storage= statement from the JobDefs without error (IIRC I got an error > message for that some time and some versions ago). > > Probably this complex configuration is the reason for your > > challenge. > > Yes, but I see no way to reduce complexity... Sadly Bareos still thinks in > "Tapes", frankly the limit of one concurrent Job per "Device" is hard to > work around... So I have to split Full, Diff and Incr for different > retention periods, and also split in different Pool sets in an attempt to > get at lease partial parallelism. > > Thanks for your help! > > Cheers, > > Martin I guess you should invest so time to read and setup spooling. I'm using that and I bundle easily 10 to 30 jobs on one media. check http://doc.bareos.org/master/html/bareos-manual-main-reference.html#QQ2-1-510 -- Bruno Friedmann Ioda-Net Sàrl www.ioda-net.ch Bareos Partner, openSUSE Member, fsfe fellowship GPG KEY : D5C9B751C4653227 irc: tigerfoot -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Hi! Am Donnerstag, 8. März 2018 01:24:33 UTC+1 schrieb Stefan Klatt: > I found only one small point. Updates generated here everytime > default files like "BackUpCatalog.conf" or "bareos-dir.conf". I > ignore the not needed files and set them to 1 Byte ("#"). We store the config in a git repository, so if some upgrade messes stuff up, reverting is easy. > > This night, I got up to 9 parallel jobs (woohoo!). Many > of them failed due to SD errors, as the number of concurrent jobs on the > director's Storages exceeds the ones on the SD (I assume it would figure this > out automatically, so I'll align them and see what happens tonight). After capping each Storage to 1 Job, I got a working backup last night, with each of the three pools sets working concurrently... Nice! Will look at the multiple-devices-per-pool concept some later day, when the need arises... Thanks again for your guidance :) Cheers, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Hallo Martin, > I have a success, see below :) That's fantastic! >> Since 16.2 you don't need the @ operator any more if you use the new >> directory structure. > Yes, the config tree stems back to good ol' Bacula 5 times... I did not get > around to "migrate" to the new bareos-dir.d style. But I do not know if > bareos does filtering based on filenames, with my '@|"sh -c 'for f in > /etc/bareos/clients/*.conf ; do echo @${f} ; done'"' I can safely ignore > editor backup/swapfiles and disable jobs by renaming them to > .conf.disabled It works like you want. I found only one small point. Updates generated here everytime default files like "BackUpCatalog.conf" or "bareos-dir.conf". I ignore the not needed files and set them to 1 Byte ("#"). >> Cleaning the JobDefs and Storage definition should make your >> configuration more clear and straight. > Indeed, apparently the nested JobDefs were the culprit. I flattened them > (Cut'n'paste all settings to the three "group" JobDefs, and removing the > top-level one). > This night, I got up to 9 parallel jobs (woohoo!). Many of them failed due to > SD errors, as the number of concurrent jobs on the director's Storages > exceeds the ones on the SD (I assume it would figure this out automatically, > so I'll align them and see what happens tonight). allowed nested JobDefs and bad behaviour or not allowed nested JobDefs without error and bad behaviour Bad bug :-( Good to know your configuration works now and you are happy with bareos Regards Stefan -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature
Re: [bareos-users] How to get parallel jobs working?
Hi! I have a success, see below :) Am Dienstag, 6. März 2018 12:05:09 UTC+1 schrieb Stefan Klatt: > Since 16.2 you don't need the @ operator any more if you use the new > directory structure. Yes, the config tree stems back to good ol' Bacula 5 times... I did not get around to "migrate" to the new bareos-dir.d style. But I do not know if bareos does filtering based on filenames, with my '@|"sh -c 'for f in /etc/bareos/clients/*.conf ; do echo @${f} ; done'"' I can safely ignore editor backup/swapfiles and disable jobs by renaming them to .conf.disabled > Cleaning the JobDefs and Storage definition should make your > configuration more clear and straight. Indeed, apparently the nested JobDefs were the culprit. I flattened them (Cut'n'paste all settings to the three "group" JobDefs, and removing the top-level one). This night, I got up to 9 parallel jobs (woohoo!). Many of them failed due to SD errors, as the number of concurrent jobs on the director's Storages exceeds the ones on the SD (I assume it would figure this out automatically, so I'll align them and see what happens tonight). Cheers, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Hello Martin, >> I have a few comments. >> >> - Do you have really monolithic config files? They are bad to read >> and old school :-). > In reality the config files are split via the @ operator (one file for pools, > one for storage, one for job templates and schedules, and one for each of ca. > 100 clients in a subdirectory)... I just reduced it for better readability > and kept only one job for each pool. Since 16.2 you don't need the @ operator any more if you use the new directory structure. > >> - For me your configuration is more complex than needed. E.g. >> Storage definitions at Pools, Jobs and JobDefs is too much. > There is no Storage definition in the Jobs. But Indeed I could remove the > Storage= statement from the JobDefs without error (IIRC I got an error > message for that some time and some versions ago). Sorry, I saw the catalog backup job has a storage definition. >> Probably this complex configuration is the reason for your >> challenge. > Yes, but I see no way to reduce complexity... Sadly Bareos still thinks in > "Tapes", frankly the limit of one concurrent Job per "Device" is hard to work > around... So I have to split Full, Diff and Incr for different retention > periods, and also split in different Pool sets in an attempt to get at lease > partial parallelism. Or more then one device per storage definition. Another way could be spooling, but I never worked with this. Cleaning the JobDefs and Storage definition should make your configuration more clear and straight. Regards, Stefan -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature
Re: [bareos-users] How to get parallel jobs working?
Hi Martin, >> I think we need to review your configuration > Apparently :) I have attached a stripped-down, sanitized example. Theoretical it should work. I have a few comments. - Do you have really monolithic config files? They are bad to read and old school :-). - I didn't know that the construct JobDefs -> JobDefs -> Job works. - For me your configuration is more complex than needed. E.g. Storage definitions at Pools, Jobs and JobDefs is too much. Probably this complex configuration is the reason for your challenge. Personally I have one file for each client (with jobs and FileSet), one minimum JobDef, for each kind of pool one file (e.g. one for file backup and one for catalog) and one file for all storage definitions. Additionally files for e.g. director and each client. I extended the new bareos configuration a little to make it more handy for me. Regards, Stefan P.s. Think simple, not so complex ;-) -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature
Re: [bareos-users] How to get parallel jobs working?
Hi! Am Sonntag, 4. März 2018 02:15:54 UTC+1 schrieb Stefan Klatt: > I think we need to review your configuration Apparently :) I have attached a stripped-down, sanitized example. Thanks! Ciao Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. Director { Name = mydc-cc001.my.domain.zz-dir QueryFile = "/usr/lib/bareos/scripts/query.sql" Maximum Concurrent Jobs = 20 Password = "Fb6f11FXwielj8EINQ934Dndoo7cRK/L2YSQUBcYlDzP" Messages = Daemon # Plugin Directory = /usr/lib64/bareos/plugins } Catalog { Name = MyCatalog dbdriver = "mysql" dbname = "bareos" dbuser = "bareos" dbpassword = "bareos" } # # Lokale konfiguration # Client { Name = mydc-cc001.my.domain.zz-fd Address = localhost FDPort = 9102 Catalog = MyCatalog Password = "w9o2LScd6JcPoZnIkAbjJC3Xm3y317vUbASuRV2UXDcKufdVkEL2QDcPvFIzXfSg" File Retention = 120 days Job Retention = 6 months AutoPrune = yes } Schedule { Name = "CatalogSchedule" Run = Full sun-sat at 23:10 } FileSet { Name = "Catalog" Include { Options { signature = MD5 } File = "/var/lib/bareos/bareos.sql" # database dump File = "/etc/bareos"# configuration } } Job { Name = "BackupCatalog" Storage = storagedaemon.my.domain.zz-backup-infrastructure Type = Backup Client = mydc-cc001.my.domain.zz-fd Messages = Standard Pool = CatalogToDiskPool Level = Full FileSet="Catalog" Schedule = "CatalogSchedule" RunBeforeJob = "/usr/lib/bareos/scripts/make_catalog_backup.pl MyCatalog" RunAfterJob = "/usr/lib/bareos/scripts/delete_catalog_backup" Write Bootstrap = "|/usr/sbin/bsmtp -h localhost -f \"\(Bareos\) \" -s \"Bootstrap for Job %j\" root@localhost" Priority = 11 } # # Standard Restore template, to be changed by Console program # Only one such job is needed for all Jobs/Clients/Storage ... # Job { Name = "RestoreFiles" Type = Restore Client = mydc-cc001.my.domain.zz-fd FileSet="RestoreSet" Pool = CatalogToDiskPool Messages = Standard Where = /tmp/bareos-restores } FileSet { Name = "RestoreSet" Include { Options { signature = MD5 } File = /usr/sbin } Exclude { File = /var/lib/bareos File = /var/lib/bareos/storage File = /proc File = /tmp File = /.journal File = /.fsck } } # When to do the backups, full backup on first sunday of the month, # differential (i.e. incremental since full) every other sunday, # and incremental backups other days Schedule { Name = "Standard" Run = Full 1st sun at 23:35 Run = Differential 2nd-5th sun at 23:35 Run = Incremental mon-sat at 23:35 } Schedule { Name = "MySQLXtraBackup" Run = Level = Full mon-sun at 23:45 } Schedule { Name = "PostgresqlLogs" Run = Level = Full sun at 1:10 Run = Level = Incremental mon-sat at 20:00 Run = Level = Incremental mon-sat at 06:00 } Schedule { Name = "PostgresqlBase" Run = Level = Full sun at 01:00 } Schedule { Name = "OracleFull" Run = Full 1st sun at 04:00 Run = Differential 2nd-5th sun at 04:00 Run = Incremental mon-sat at 04:00 } Schedule { Name = "Daytime" Run = Full 1st sun at 10:00 Run = Differential 2nd-5th sun at 10:00 Run = Incremental mon-sat at 10:00 } Messages { Name = Standard mailcommand = "/usr/sbin/bsmtp -h localhost -f \"\(Bareos\) \<%r\>\" -s \"Bareos: %t %e of %c %l\" %r" operatorcommand = "/usr/sbin/bsmtp -h localhost -f \"\(Bareos\) \<%r\>\" -s \"Bareos: Intervention needed for %j\" %r" mail = xxx = all, !skipped operator = xxx = mount console = all, !skipped, !saved syslog = all catalog = all } Messages { Name = Daemon mailcommand = "/usr/sbin/bsmtp -h localhost -f \"\(Bareos\) \<%r\>\" -s \"Bareos daemon message\" %r" mail = = all, !skipped console = all, !skipped, !saved append = "/var/log/bareos/bareos.log" = all, !skipped } # # Verwaltungszugänge: Console (BAT, etc.) sowie Web-UI # Console { Name = mydc-cc001.my.domain.zz-mon Password = "xV5mv0r8i2mMlZLShJFNwU50r2dZZS1j3NRh2U68ETIa" CommandACL = status, .status } Console { Name = admin Password = "b!Happ/" Profile = webui } Profile { Name = webui CommandACL = !.bvfs_clear_cache, !.exit, !.sql, !configure, !create, !delete, !purge, !sqlquery, !umount, !unmount, *all* Job ACL = *all* Schedule ACL = *all* Catalog ACL = *all* Pool ACL = *all* Storage ACL = *all* Client ACL = *all* FileSet ACL = *all* Where ACL = *all* Plugin Options ACL = *all* } FileSet { Name = "mydc-cc001.my.domain.zz-Set" Include { Options { signature = MD5 onefs = no aclsupport = no sparse = yes
Re: [bareos-users] How to get parallel jobs working?
Hello Martin, >> I might have a clue, I had a "Pool" and "Storage" statement in the top-level >> jobDefs (although I use "Full Backup Pool", "Differential Backup Pool" etc. >> to set the final target pool. >> Maybe that throws the scheduler off, I removed Pool and Storage from the >> top-level jobDefs, will try if it helps... > Hmm, did not help, still no parallel jobs. > > to recap, I have: > > On the director: > * Director Maximum Concurrent Jobs = 20 > * 3 Storages, each Storage: Maximum Concurrent Jobs = 20 > > On the storage daemon: > * Storage: Maximum Concurrent Jobs = 10 > * 3 Devices, each Device: Maximum Concurrent Jobs = 4 (Should be irrelevant, > as each device can serve only one "drive" with one volume simultaneously, > anyways. > > So I expect that each device should be able to process one Volume, thus one > Job, independently, allowing 3 jobs (one for each pool set) to run in > parallel. I think we need to review your configuration > or am I missing something? Probably... or a bug... Regards, Stefan -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature
Re: [bareos-users] How to get parallel jobs working?
Am Mittwoch, 28. Februar 2018 15:19:33 UTC+1 schrieb Martin Emrich: > I might have a clue, I had a "Pool" and "Storage" statement in the top-level > jobDefs (although I use "Full Backup Pool", "Differential Backup Pool" etc. > to set the final target pool. > Maybe that throws the scheduler off, I removed Pool and Storage from the > top-level jobDefs, will try if it helps... Hmm, did not help, still no parallel jobs. to recap, I have: On the director: * Director Maximum Concurrent Jobs = 20 * 3 Storages, each Storage: Maximum Concurrent Jobs = 20 On the storage daemon: * Storage: Maximum Concurrent Jobs = 10 * 3 Devices, each Device: Maximum Concurrent Jobs = 4 (Should be irrelevant, as each device can serve only one "drive" with one volume simultaneously, anyways. So I expect that each device should be able to process one Volume, thus one Job, independently, allowing 3 jobs (one for each pool set) to run in parallel. or am I missing something? Cheers, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Am Dienstag, 27. Februar 2018 21:16:12 UTC+1 schrieb Stefan Klatt: > > Do you mean with director the job definitions? > > Do you use different job definitions? If not do you use "Allow > Duplicate Jobs"? > I have one master "jobDefs". Then I have tree for the three Pool sets, the final client jobs in turn inherit from them. I might have a clue, I had a "Pool" and "Storage" statement in the top-level jobDefs (although I use "Full Backup Pool", "Differential Backup Pool" etc. to set the final target pool. Maybe that throws the scheduler off, I removed Pool and Storage from the top-level jobDefs, will try if it helps... Cheers, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Hello Martin >> Probably a problem with the job priority? > Hmm can you elaborate? I have "Allow Mixed Priority" already set to yes > (tried the default "no" before, too). > Otherwise, all Jobs have the same priority 10. Do you mean with director the job definitions? Do you use different job definitions? If not do you use "Allow Duplicate Jobs"? Regards, Stefan -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature
Re: [bareos-users] How to get parallel jobs working?
Hi Stefan! Am Montag, 26. Februar 2018 19:20:21 UTC+1 schrieb Stefan Klatt: > > Probably a problem with the job priority? Hmm can you elaborate? I have "Allow Mixed Priority" already set to yes (tried the default "no" before, too). Otherwise, all Jobs have the same priority 10. Thanks, Martin -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] How to get parallel jobs working?
Hi Martin, Am 26.02.2018 um 15:43 schrieb Martin Emrich: > Hi! > > I am trying to get parallel jobs working, but it just does not work: > > * I have separate pools, on separate devices > * I have set "maximum concurrent jobs" set on the director (dir and storages) > and on the storage (storage and devices) > > But when I start three jobs (each configured for a different pool), the other > two hang with "waiting for execution". > > What did I miss? > > Thanks so much > > Martin > Probably a problem with the job priority? Regards, Stefan -- *CaC, Computer and Communication* Inhaber Stefan Klatt End-2-End Senior Network Consultant CISSP / CISM / ISO27001 Lead Implementer / TOGAF9 Badges: https://www.youracclaim.com/users/stefan-klatt Triftstrasse 9 60528 Frankfurt Germany USt-IdNr.: DE260461592 Tel.: +49-(0)172-6807809 Tel.: +49-(0)69-67808-900 Fax: +49-(0)69-67808-837 Email: stefan.kl...@cac-netzwerk.de Profil: http://www.cac-netzwerk.de/profil -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout. smime.p7s Description: S/MIME Cryptographic Signature