Re: [Bacula-users] verify job fails, must init catalog run twice?
> to me it looks like InitCatalog must run twice before all changes are known > and the regular verify behaves the way it should. > > (note: I have not done any changes to the system between the two jobs and > auto update is disabled. I am not running a virus scanner or anything else > that could touch the files). one more addition: the "special" thing on my setup is that its using *two* catalogs. One for backup jobs, one for verify jobs. Could that confuse bacula? Stephan - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] verify job fails, must init catalog run twice?
Hello, I see this for quite a while now and it does not seem to disappear with 2.0.1... maybe someone has the same problem? yesterday I was upgrading my system and ran a "InitCatalog" manually thereafter. This morning my regular verify job failed with reporting that some hundred files have changed (all in /lib, /sbin and /usr - stuff updated by rpm). 2007-03-13 08:37:11 bender.in.oldnewyork: verify-bender.2007-03-13_08.30.00 [...] Error: Bacula 2.0.1 (12Jan07): 13-Mar-2007 08:37:11 JobId: 183 Job: verify-bender.2007-03-13_08.30.00 FileSet: verify-bender Verify Level: Catalog Client: bender-verify Verify JobId: 182 Verify Job: Start time: 13-Mar-2007 08:30:02 End time: 13-Mar-2007 08:37:11 Files Examined: 49,803 Non-fatal FD errors: 0 FD termination status: OK Termination: *** Verify Error *** now I repeated the very same procedure: manual "InitCatalog" job, then the regular (but this time manually started) verify job. This time it went OK. No differences found. I bet that tomorrows automatic verify job will also terminate OK as that is what always happened previously. 13-Mar 11:45 bender.in.oldnewyork: Bacula 2.0.1 (12Jan07): 13-Mar-2007 11:45:19 JobId: 185 Job:verify-bender.2007-03-13_11.38.07 FileSet:verify-bender Verify Level: Catalog Client: bender-verify Verify JobId: 184 Verify Job: Start time: 13-Mar-2007 11:38:09 End time: 13-Mar-2007 11:45:19 Files Examined: 49,803 Non-fatal FD errors:0 FD termination status: OK Termination:Verify OK to me it looks like InitCatalog must run twice before all changes are known and the regular verify behaves the way it should. (note: I have not done any changes to the system between the two jobs and auto update is disabled. I am not running a virus scanner or anything else that could touch the files). regards, Stephan PS: i can post the complete bconsole conversations and/or config if needed - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] [Bacula-devel] Misc
Hello, On Thursday 01 March 2007 10:32:49 Kern Sibbald wrote: [...] > 2. Doing a backup/restore from/to a FIFO (as defined in a FileSet resource) > is supported and from my tests works perfectly well in Bacula 2.0.x. In > some previous versions there were some problems, but they have long since > been corrected. this item does indeed work very stable. I've been using this in production for exporting large Oracle databases (100+ GBytes each) once per week for about two years. I've never encountered stability issue with the FIFOS. :-) There was only one issue during restores (not a stability one though): the FIFO permissions were -rw--- and owned by root. Therefore the 'oracle' user was not able to read it for importing... though, I do not remember whether the reason was bacula or my own scripts?! Anyhow, bacula does not seem to wait long when there is nothing reading from the FIFO. So one must be quick with fixing the permissions... all my saying goes for 1.38.11 and previous versions. Unfortunately I can not tell for 2.0.x since I no longer have control over this installation... regards, Stephan - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] [Bacula-devel] FOSDEM
On Tuesday 27 February 2007 19:02:20 Kern Sibbald wrote: > On Tuesday 27 February 2007 18:18, Eric Bollengier wrote: > > if you want to see kern in action ! > > > > http://ftp.belnet.be/mirrors/FOSDEM/2007/FOSDEM2007-Bacula.ogg > > Egads, what a horror! :-) > > Every time I do a live demo, something goes wrong. What I forgot was that > in a certain sense, my whole presentation was a live demo because I had to > run OpenOffice to display my slides -- it started off very slowly with the > SuSE update consuming my whole underpowered machine for about 5 minutes :-( there was nothing wrong. I think you just showed very live and clearly how good bacula and OSS in general is. ;-) What I mean is: SuSE's ZLM + KDE + OpenOffice + apache (with mod_perl) + Firefox + bacula (dir, fd, sd, bconsole) + Konsoles + perhaps some other KDE applets and daemons... all on a 256 MByte/700MHz machine with slow IDE disk. That is fantastic IMO. :-) Stephan - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Long term data archiving - howto?
Hello, On Thursday 15 February 2007 11:16:06 > Date: Wed, 14 Feb 2007 12:34:34 -0800 > Hi, > > I would like to archive some data to tape and keep it around forever. > Would using bacula to do this be the right way? Or would simply > tarring them up to the tape be better? > > This is a one time job (and not a regular backup), and I have defined > a job for ad-hoc stuff like this within bacula. I can set the volume > (that this job will go to) to have a very long volume retention, but > that still leaves the job records. I use a separate Pool for this. With very long retention periods, as you say. Then I make sure bacula does never attempt to recycle volumes in that pool (Recycle = no). Also you may set AutoPrune=no which should keep job records forever (unless you manually prune volumes). On the other hand, you may even loose the job records. Its easy to re-create them by reading the volume back in with bscan -m. Might be time consuming depending on the volume size/speed but works very well. The important point is to not recycle the volume(s) by accident. For tapes: just flip the read-only switch... :-) Stephan - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet quirks
Hello, On Tuesday 06 February 2007 20:28, Martin Simmons wrote: > > regexdir = "/home/.*/[^Movies|^Music|^Pictures]/.*" > > > > (though, this doesnt work yet) > > No, you can't use [] like that in a regexp -- it is for matching single > characters, not strings. There is no "not" operator in regexps. there is. Ie this works egrep "/home/.*/[^Pictures]" pathlist.file the only problem is it would also _not_ match /home/user/iPctures because of the nature of []... I have not tried this with bregex because I prefer your solution below. > > Somehow I am stuck here. Can someone see a more simple approach to get > > _only_ those three directories from each home in the system? Or do I have > > to write a script? > > This does most of what you want: > > FileSet { > Name = "data-users-media" > > Include { > > Options { > signature = md5 > wilddir = "/home/*/Movies" > wild = "/home/*/Movies/*" > wilddir = "/home/*/Music" > wild = "/home/*/Music/*" > wilddir = "/home/*/Pictures" > wild = "/home/*/Pictures/*" > } > > Options { > wild = "/home/*/*" > Exclude = yes > } > > File = /home > } > } this does the trick. Thanks a lot. > The only problem is that it also includes everything directly in /home, but > that may not matter. not that I can see. The wild = "/home/*/*" directive also seems to match "/home/*" for a reason that I do not quite understand. Adding wilddir = "/home/*" to the include block selects all dirs directly in /home, which indicates that everything was selected by the File = /home directive... also removing the exclude block selects everything (both tested with estimate). So "/home/*" must be dropped in the "/home/*/*" match IMHO. Stephan - 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
[Bacula-users] FileSet quirks
Hello, back in 1.38.11 I wrote this fileset which appeared to work quite nice. FileSet { Name = "data-users-media" Include { Options { signature = md5 wilddir = "/home/*/Movies" wilddir = "/home/*/Music" wilddir = "/home/*/Pictures" } # I do not want backup files Options { wildfile = "*~" Exclude = yes } } } Now, in bacula 2.0.1 it does not work anymore. Zero directories are selected. Apparently this is because its missing a 'File = /home' directive in the Include block. Adding this will select *all* files/dirs in /home which means I must somehow exclude everything except the three directories I want. But I do not know what to exclude exactly. There is just too much stuff in multiple homes on multiple computers... so I try: FileSet { Name = "data-users-media" Include { Options { signature = md5 wilddir = "/home/*/Movies" wilddir = "/home/*/Music" wilddir = "/home/*/Pictures" } Options { wild = "/home*" Exclude = yes } File = /home } } (omitting the backup files for now) no luck. Only /home itself is selected. Nothing underneath. Also experimenting with wilddir and wildfile in the exclude block does not take me any further. Now I am trying to write a more-or-less complicated regexp to revert-match the three dirs in the Exclude block at once. Something like: regexdir = "/home/.*/[^Movies|^Music|^Pictures]/.*" (though, this doesnt work yet) Somehow I am stuck here. Can someone see a more simple approach to get _only_ those three directories from each home in the system? Or do I have to write a script? thanks, Stephan - 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
Re: [Bacula-users] can not write to DVD
Hello, > > I am using 'UDF formatted DVD-RAMs' for backup at home. So it's the > > write-files-to-disk approach rather than the write-data-part-to-media > > approach. > > > > To mount the media automatically when bacula needs it (and umount after > > backup) I am using autofs. [...] > > I can post the config if anyone is interested. Just let me know. > > Yes, Please I would love to see how you are doing this. > no problem. Autofs needs two config files: /etc/auto.master and /etc/auto.dvd-ram. The first one is mandatory. It defines where autofs maintains those 'special' mount points. The later one is created by me - it contains all settings for the actual mount point. Here are the essential lines: [EMAIL PROTECTED] ~]# cat /etc/auto.master [...] /mnt/autofs/etc/auto.dvd-ram --timout=600 --ghost and [EMAIL PROTECTED] ~]# cat /etc/auto.dvd-ram [...] dvdwriter -fstype=udf,rw,nosuid,nodev,noatime,user :/dev/dvdwriter the timeout in auto.master defines after which (idle) time autofs should try to umount. Maybe it is better so set a somewhat larger value - I don't know what happens when autofs tries to umount while there is still a bacula job running (and waiting for something)... Testing it: provided the autofs daemon is running and there is a UDF formated DVD-RAM in the drive you shold be able to do this: [EMAIL PROTECTED] ~]# ls -l /mnt/autofs/dvdwriter/ < --- this takes a couple seconds > total 173156 -rw-r- 1 bacula disk 177311477 Aug 29 19:45 Daily-0005 drwxr-xr-x 2 root root40 Jun 11 21:30 lost+found autofs should 'block' the ls command while mounting the DVD seamlessly. You notice a more-or-less short delay. If ls returns right away and you get a empty directory listing then something doesn't work. Check the usual log files (ie /var/log/messages). Finally here is my device resource operating on that mount point. It's essentially a regular 'disk' device Device { Name = DVD-Writer Media Type = DVD DeviceType = File # autofs will deal with mounting # - see /etc/auto.master and /etc/auto.dvd-ram ArchiveDevice = /mnt/autofs/dvdwriter MountPoint = /mnt/autofs/dvdwriter # general parameters LabelMedia = no AutomaticMount = yes RemovableMedia = yes AlwaysOpen = no } regards, Stephan - 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
[Bacula-users] udev rules for bacula (was: Re: autochanger device has 0 slots)
Hello, (I post this again, since my first post a couple days ago did not reach the list for some reason.) On Monday 07 August 2006 14:01, Kern Sibbald wrote: > > Is there a chance to get udev rules shipped with the Fedora RPM Packages? > > Yes, this is probably a good idea because there are probably very few users > who know how to create udev rules. I'll look at adding them to the > release, but I'm not sure we will install them on the system, but rather > leave the file in /etc/bacula for the user to install should he wish. ok, here is what works for me (sorry for the delay). The basic Idea of my rules is to first fix the permissions on the real device node(s), so that bacula works. Then, optionally, create persistent symlinks for use in the bacula-sd.conf file. It should be possible to shuffle SCSI hardware around without worries about breaking the bacula configuration (ie volumes loaded in the wrong drive since nst0 and nst1 have switched, etc...) The rules goes into /etc/udev/rules.d/55-bacula.rules (at least in Fedora, other distros might differ) I put in one examples to create the persistent symlinks. It is matching the devices by the output of the scsi_id program. I use this here in production. All that depends strongly on the hardware used, so it is commented out in the attached file. To got things going, it is necessary to adjust the strings in RESULT. The rest *should* work quite distribution independent. Hopefully. To get a RESULT string for your autochanger run something like this (provided the changer is currently sg5): scsi_id -g -u -s /class/scsi_generic/sg5 for a tape drive the command would be: scsi_id -g -u -s /class/scsi_tape/nst0 whatever string those commands return, put it in the udev config file. If the matching works you should see device links like that in /dev: # ls -la /dev/autochanger1* lrwxrwxrwx 1 root root 3 Aug 15 20:00 /dev/autochanger1 -> sg5 lrwxrwxrwx 1 root root 4 Aug 15 20:00 /dev/autochanger1-drive1 -> nst1 lrwxrwxrwx 1 root root 4 Aug 15 19:59 /dev/autochanger1-drive2 -> nst0 (that is only my philosophy of representing one changer with two drives, it can be named however you like) The links should remain stable, while their targets (sg5, nst1, nst0) might change if something changes on the SCSI bus. best regards, Stephan # # Setup Autochanger Devices for use with bacula (http://www.bacula.org) # # - tested on Fedora Core 4 with bacula-mysql RPM package # # fix the permissions of the standard devices # - this is enough to get bacula going KERNEL=="sg[0-9]*", BUS=="scsi", MODE="0660", GROUP="disk" # optional: # - create persistent symlinks for use in bacula-sd.conf # - adjust the RESULT strings to match your specific hardware # (you can run the scsi_id program on the command line to get correct strings) # #SUBSYSTEM=="scsi_generic", PROGRAM=="/sbin/scsi_id -g -d %N -s %p", RESULT=="200900de794345a07", SYMLINK+="autochanger1" #KERNEL=="nst*", SUBSYSTEM=="scsi_tape", PROGRAM=="/sbin/scsi_id -g -u -d %N -s %p" RESULT=="1IBM_ULTRIUM-TD2_1110279276" SYMLINK+="autochanger1-drive1" #KERNEL=="nst*", SUBSYSTEM=="scsi_tape", PROGRAM=="/sbin/scsi_id -g -u -d %N -s %p" RESULT=="1IBM_ULTRIUM-TD2_1110279611" SYMLINK+="autochanger1-drive2" - 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
Re: [Bacula-users] can not write to DVD
Hello, > > /dev/hdc /media/cdrom1 iso9660 ro,user,noauto 0 0 > > > > to /etc/fstab > > Yes, my /etc/fstab looks like this! I tried manually mount and I did a > reboot. This is not helping :( > > > I must admit that to-date I've run as root for DVD stuff. That's > > something to think about once we've got the 1.39 DVD stuff working. > > Should I run bacula-sd as root? > > What else should I try to get the DVD mounted? I am not sure whether the following fits to your problem (I did not follow this thread). If it doesn't help, please ignore my post: I am using 'UDF formatted DVD-RAMs' for backup at home. So it's the write-files-to-disk approach rather than the write-data-part-to-media approach. To mount the media automatically when bacula needs it (and umount after backup) I am using autofs. That is a daemon which does the 'mount/umount' part for you (or bacula for that matter). It requires similar informations as in fstab. Then it listens on the mount point through some kernel mechanism. Whenever a program (ie bacula-sd) is trying to access the directory, autofs will kick in and perform the mount operation as per configuration. The accessing program (bacula-sd) is just put on hold for as long as it takes. Once mounted autofs will keep on watching the mount point. When it is idle for a certain amount of time, it will just umount it. The cool thing is that I can always change DVDs without doing anything but pressing the eject button on the drive (it's locked when mounted). No additional scripting and/or RunBefore/After jobs needed in bacula. I can post the config if anyone is interested. Just let me know. best regards, Stephan - 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
[Bacula-users] udev rules for bacula [was: Re: autochanger device has 0 slots]
Hello, On Monday 07 August 2006 14:01, Kern Sibbald wrote: > > Is there a chance to get udev rules shipped with the Fedora RPM Packages? > > Yes, this is probably a good idea because there are probably very few users > who know how to create udev rules. I'll look at adding them to the > release, but I'm not sure we will install them on the system, but rather > leave the file in /etc/bacula for the user to install should he wish. ok, here is what works for me (sorry for the delay). The basic Idea of my rules is to first fix the permissions on the real device node(s), so that bacula works. Then, optionally, create persistent symlinks for use in the bacula-sd.conf file. It should be possible to shuffle SCSI hardware around without worries about breaking the bacula configuration (ie volumes loaded in the wrong drive since nst0 and nst1 have switched, etc...) The rules goes into /etc/udev/rules.d/55-bacula.rules (at least in Fedora, other distros might differ) I put in one examples to create the persistent symlinks. It is matching the devices by the output of the scsi_id program. I use this here in production. All that depends strongly on the hardware used, so it is commented out in the attached file. To got things going, it is necessary to adjust the strings in RESULT. The rest *should* work quite distribution independent. Hopefully. To get a RESULT string for your autochanger run something like this (provided the changer is currently sg5): scsi_id -g -u -s /class/scsi_generic/sg5 for a tape drive the command would be: scsi_id -g -u -s /class/scsi_tape/nst0 whatever string those commands return, put it in the udev config file. If the matching works you should see device links like that in /dev: # ls -la /dev/autochanger1* lrwxrwxrwx 1 root root 3 Aug 15 20:00 /dev/autochanger1 -> sg5 lrwxrwxrwx 1 root root 4 Aug 15 20:00 /dev/autochanger1-drive1 -> nst1 lrwxrwxrwx 1 root root 4 Aug 15 19:59 /dev/autochanger1-drive2 -> nst0 (that is only my philosophy of representing one changer with two drives, it can be named however you like) The links should remain stable, while their targets (sg5, nst1, nst0) might change if something changes on the SCSI bus. best regards, Stephan # # Setup Autochanger Devices for use with bacula (http://www.bacula.org) # # - tested on Fedora Core 4 with bacula-mysql RPM package # # fix the permissions of the standard devices # - this is enough to get bacula going KERNEL=="sg[0-9]*", BUS=="scsi", MODE="0660", GROUP="disk" # optional: # - create persistent symlinks for use in bacula-sd.conf # - adjust the RESULT strings to match your specific hardware # (you can run the scsi_id program on the command line to get correct strings) # #SUBSYSTEM=="scsi_generic", PROGRAM=="/sbin/scsi_id -g -d %N -s %p", RESULT=="200900de794345a07", SYMLINK+="autochanger1" #KERNEL=="nst*", SUBSYSTEM=="scsi_tape", PROGRAM=="/sbin/scsi_id -g -u -d %N -s %p" RESULT=="1IBM_ULTRIUM-TD2_1110279276" SYMLINK+="autochanger1-drive1" #KERNEL=="nst*", SUBSYSTEM=="scsi_tape", PROGRAM=="/sbin/scsi_id -g -u -d %N -s %p" RESULT=="1IBM_ULTRIUM-TD2_1110279611" SYMLINK+="autochanger1-drive2" - 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
Re: [Bacula-users] autochanger device has 0 slots
On Thursday 03 August 2006 21:14, Timo Neuvonen wrote: > > Now the 'update slots' command does not recognize the > > number of slots in my autochanger [1] anymore. > > Check the permissions for your changer device (/dev/sg5). > > Does storage daemon have rightsto access it? If SD is running as > bacula:disk as I expect, sg5 should have group disk with group privileges > rw-. I think FC4 used to set this to something more restricted by default. > And in 1.38.0 SD propably was running as root by default, so that would > explain the problem. > > If this is the problem make sure the permissions retain the correct values > after reboot: you'll propably need to modify something in /etc/udev, since > results of chmod / chown will be "overwritten" during the next boot. ah, thanks. That was problem #1. FC4 has --- group privileges by default. I could not tweak udev correctly yet. It sets the permissions, but not the group for some reason. Also I learned that it is 'incorrect' to modify /etc/udev/rules.d/50-udev.rules. The correct way is to create a separate file with a lower number in the file name. Ie '40-bacula.rules'. I will post one as soon as I get it working. Is there a chance to get udev rules shipped with the Fedora RPM Packages? Problem #2 was that 'mtx' is located in /usr/sbin, which is not in the default PATH of the 'bacula' user. I fixed that straight in mtx-changer for now. Problem #3 were the privileges of mtx-changer itself. That was likely because of me playing around... it did a lot FULLs this weekend with a lot of tape switching. Looks all good now. thanks for the help, Stephan - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] autochanger device has 0 slots
Hello, I upgraded from 1.38.0 to 1.38.11 (28 June 2006), all running on FC4. Now the 'update slots' command does not recognize the number of slots in my autochanger [1] anymore. Here is the console conversation: *update slots storage=tape1 Enter autochanger drive[0]: 0 Connecting to Storage daemon tape1 at dd-lx-backup.dd.net-linx:9103 ... 3306 Issuing autochanger "slots" command. Device "changer1" has 0 slots. No slots in changer to scan. the mtx-changer script however, seems to work as expected: # /etc/bacula/mtx-changer /dev/sg5 slots 0 /dev/nst0 0 30 # # /etc/bacula/mtx-changer /dev/sg5 list 0 /dev/nst0 0 1:001154L2 2:001153L2 3:001152L2 [...] 27:001157L2 28:001156L2 # I read somewhere in the manual manual (or list?) that the 'slots' command should actually return the number of slots - *not* print it. The released mtx-changer script, however, *does* print this number. I modified it to return the number of slots but that did not have any effect on the 'update slots' console command. Am I doing something wrong here? thanks and regards, Stephan [1] the autochanger setup is as follows: Autochanger { Name= "changer1" Device = tape1, tape2 Changer Device = /dev/sg5 # why 5? Don't know... Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d" } Device { Name= tape1 Autochanger = yes Media Type = LTO Archive Device = /dev/nst0 Drive Index = 0 Removable Media = yes Automatic Mount = yes Always Open = yes Spool Directory= /u01/bacula-spool Maximum Job Spool Size = 1000 } Device { Name= tape2 Autochanger = yes Media Type = LTO Archive Device = /dev/nst1 Drive Index = 1 Automatic Mount = yes Always Open = yes Removable Media = yes Spool Directory= /u01/bacula-spool Maximum Job Spool Size = 1000 } - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] subversion backup at SourceForge with public FD?
Alan Brown wrote: > On Thu, 27 Apr 2006, Stephan Ebelt wrote: > >> However, from the implementation point of view I think this is a bit of >> overhead since it requires local disk space being available and there >> must be additional logic to detect whether rsync actually succeeded... >> kind of similar to dumping some large SQL database (where one often can >> not afford the temporary disk space) > > In the specific case of SQL database backup, if you halt the SQL server > then you can safely backup the databases, however a dump means the > tables can be reconstructed on almost any SQL-speaking database server > (modulo minor tweaks) yes. I just wanted another example where one might need additional temporary disk space during the backup process. SQL dumps just came in mind because of the way we use oracle here (it is not exactly the 'standard' way). We got around the 'temporary disk space' problem by pushing the dump straight through FIFOs into bacula (which works pretty well by the way). I can give more details off-list if someone is interested. best regards, Stephan --- 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
[Bacula-users] subversion backup at SourceForge with public FD?
Hello, I just managed to convert my own project over to subversion and do now think about backup (I run bacula at home for various other reasons and want it to backup my project data as well). SourceForge provides a public rsync service. So its no problem to make a RunBeforeJob to update a local copy of the repository and then let bacula store it. However, from the implementation point of view I think this is a bit of overhead since it requires local disk space being available and there must be additional logic to detect whether rsync actually succeeded... kind of similar to dumping some large SQL database (where one often can not afford the temporary disk space) So I wonder how cool it would be to have a public FD running at SourceForge that I (or anyone else) could use directly from a Director at home. I could run jobs as usual. Without the need of an 'rsync layer' in between. Would the FD be sufficient for such a task? (ie. it must probably run a fairly large number of jobs at the same time?) And as a consequence would it be worth suggesting to SF? (I would not do that myself as I can not speak for the bacula project) ... just a thought ... best regards, Stephan --- 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
[Bacula-users] Re: [Bacula-devel] Proposed enhancement to Bacula-Web
Kern Sibbald wrote: > Hello, > > I'm a bit surprised because I have not seen any comments on this proposal. Am > I missing something? It sounds like a nice feature to me. to me too. It could be useful e.g to show somebody how much work the backup system actually does. Ie during the last week. If its in the CVS I would be willing to put the missing form in (if Mike has not done so yet). I would prefer the input being a 'select' list with human readable terms such as '2 days ago', '3 days ago', 'one week ago', ..., 'half year ago', ... for the lazy folks who like to click. Actually we could also do both. :-) best regards, Stephan --- This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] RPMs for RHEL4 on x86_64, mysql
Hello, I just build those packages from bacula-1.38.4-1.src.rpm. It took me (as a non-c-programmer-and-rpmbuild-newbe) less than half an hour. Inclusive installing all dependencies before building. The build process itself worked very smooth. Then I could roll out the packages to all my shiny new servers. I am impressed. (especially when comparing that to the usual software installation nightmare on certain systems from Redmond) Many thanks to everyone working on bacula! If anyone is interested in these packages - just let me know. I can also build postgresql if needed. One question: I havn't seen any x86_64 packages in the file releases list yet. Is there a plan to do that for upcoming releases? Or would that be something to contribute? best regards, Stephan --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642 ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet accross mountpoints with regular expressions
Hello, Kern Sibbald wrote: [...] That said, once you prooved that your fileset works, it would be a good contribution to the manual. I agee with Arno. Your approach is the best one I have seen, and would be a nice contribution to the manual. last night all went just fine with this FileSet. No warnings, all NFS mounts were skipped and it made Incrementals just as desired. is that proof enough? I don't know what else to say as it just works. :-) best regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet accross mountpoints with regular expressions
Kern Sibbald wrote: That said, once you prooved that your fileset works, it would be a good contribution to the manual. I agee with Arno. Your approach is the best one I have seen, and would be a nice contribution to the manual. ok, last night all jobs ran and it looks good. All finished with 'Backup OK'. No warnings. One client has NFS mounts below /u02 and I get messages like this: [...] dd-lx-fs01.dd.net-linx: Disallowed filesystem. Will not descend into /u02/oraout/dd-lx-oracle4 dd-lx-fs01.dd.net-linx: Disallowed filesystem. Will not descend into /u02/oraout/Trace/DD11 dd-lx-fs01.dd.net-linx: Disallowed filesystem. Will not descend into /u02/oraout/Trace/DD41 [...] the restore command does not list anything below these directories. So I would say it works. * restore [...] $ cd /u02/oraout/Trace/DD41/ cwd is: /u02/oraout/Trace/DD41/ $ dir Node DD41 has no children. $ . one thing I am a bit concerned about (but shouldn't have anything to do with the REs?): it upgraded all jobs from Incremental to Full despite the IgnoreFileSetChanges option. I havn't seen the usual 'fileset was modified - performing Full backup' message anywhere. All job reports say: 09-Nov 23:27 dd-lx-backup.dd.net-linx: No prior Full backup Job record found. 09-Nov 23:27 dd-lx-backup.dd.net-linx: No prior or suitable Full backup found. Doing FULL backup. 09-Nov 23:27 dd-lx-backup.dd.net-linx: Start Backup JobId 2028, Job=dd-lx-www.2005-11-09_22.00.07 09-Nov 23:27 dd-lx-backup.dd.net-linx: Volume "disk3-0008" previously written, moving to end of data. 09-Nov 23:28 dd-lx-backup.dd.net-linx: Bacula 1.38.0 (28Oct05): 09-Nov-2005 23:28:06 JobId: 2028 Job:dd-lx-www.2005-11-09_22.00.07 Backup Level: Full (upgraded from Incremental) Client: "dd-lx-www" i386-redhat-linux-gnu,redhat,7.3 FileSet:"FileServer" 2005-11-08 15:57:14 [...] but there was definitly a Full available. Even more than one. Here's a list of jobs from the last couple days (I cut off the JobFiles and JobBytes columns - let me know when you need that): | JobId | Client | FileSetLevel | StartTime| VolumeName | [...] | 1846 | dd-lx-www | FileServer | D | 2005-10-31 00:05:20 |disk3-0026 | | 1868 | dd-lx-www | FileServer | I | 2005-10-31 23:23:27 |disk3-0028 | | 1889 | dd-lx-www | FileServer | I | 2005-11-01 23:22:24 |disk3-0029 | | 1909 | dd-lx-www | FileServer | I | 2005-11-02 23:29:27 |disk3-0006 | | 1929 | dd-lx-www | FileServer | I | 2005-11-03 23:28:24 |disk3-0030 | | 1949 | dd-lx-www | FileServer | F | 2005-11-05 00:16:28 |disk3-0009 | | 1967 | dd-lx-www | FileServer | I | 2005-11-06 03:20:21 |disk3-0031 | | 1984 | dd-lx-www | FileServer | F | 2005-11-07 12:16:23 |disk3-0011 | | 2006 | dd-lx-www | FileServer | I | 2005-11-07 23:16:25 |disk3-0007 | | 2028 | dd-lx-www | FileServer | F | 2005-11-09 23:27:24 |disk3-0008 | All retention periods are set to 30 days. the FileSet table contains a new record: +--+++-+ |FileSetId | FileSet| MD5| CreateTime | +--+++-+ |4 | FileServer | n8/0H5p2IzspQ/U/s/+qrC | 2005-08-05 22:00:03 | [...] | 12 | FileServer | p8+4OD/V6Q+P3i883QZ4NC | 2005-11-08 15:57:14 | +--+++-+ best regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet accross mountpoints with regular expressions
Stephan Ebelt wrote: [...] The job didn't run because bacula seems to have crashed. I've got no glue why yet. All I found till now is - no director process was running this morning - none of the scheduled jobs did run as I've got not a single e-mail - the message below appears in the .conmsg file as well as in the last written .mail file in /var/bacula. (both files are timestamped in the same second) 08-Nov 19:00 dd-lx-backup.dd.net-linx: Fatal Error because: Bacula interrupted by signal 11: Segmentation violation all right, after browsing the devel list and bug system I would suspect this is the same as reported as bug #375 'Resched failure after configuration change and reload'. I've done plenty of reloads yesterday while testing the FileSet. Unfortunately I do not have any trace. Or I can't find it. It is an rpm installation were debugging is probably disabled? gdb was not installed. It is now and I should be able to produce another trace if needed. best regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet accross mountpoints with regular expressions
Hello. Kern Sibbald wrote: Quite possible, considering the number of possible ways to use options blocks in filesets. I fact I suspect that Kern himself doesn't fully understand the implications of options with includes and exludes and REs and whatnotelse ;-) Yes, Arno is correct :-) does that mean bacula has some sort of 'artificial intelligence' build in? That's the killer feature over all other backup systems. ;-) That said, once you prooved that your fileset works, it would be a good contribution to the manual. I agee with Arno. Your approach is the best one I have seen, and would be a nice contribution to the manual. thanks. there's one good news and one bad so far. The backup job from last night did not run, so it doesn't proof that the FileSet is /not/ working. ;-) The job didn't run because bacula seems to have crashed. I've got no glue why yet. All I found till now is - no director process was running this morning - none of the scheduled jobs did run as I've got not a single e-mail - the message below appears in the .conmsg file as well as in the last written .mail file in /var/bacula. (both files are timestamped in the same second) 08-Nov 19:00 dd-lx-backup.dd.net-linx: Fatal Error because: Bacula interrupted by signal 11: Segmentation violation Unfortunately I do not have any trace. Or I can't find it. It is an rpm installation were debugging is probably disabled? I still have the *.state files. Not sure if they're interesting. They're timestamped ~2 hours before the message appeared. I will restart bacula now and see. I'll let you know if that happens again or when I get more 'evidence'. best regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] FileSet accross mountpoints with regular expressions
Hello, Arno Lehmann wrote: SITUATION There are multiple clients and I like to have one fileset for all of them. As they are quite similar. Each one looks like /u01 /u02 ... /u09 /u10 ... /uNN Interesting setup... it is actually what oracle recommends to mount data disks. Not sure if they still do, but a while ago we just took that and made it a internal standard. To get server installations looking more alike. The idea behind is that everything in /uNN dirs is data while everything outside of them is OS stuff, log files, applications, ... the (more or less) standard unix directories so to say... this way you always know what you better not wipe out. ;-) approach as it would IMHO cause me a headache to write a shell command that works on all my OSes with different shell/find/awk/... versions. Well, with different OSes and shells and tools that would be an interesting exercise. Or, in other words, I understand why you don't want that way. if you like this kind of exercise, here's what we've got: ;-) - RedHat 7, 8, 9 - Fedora 2, 4 - SuSE 6.4(!), 9, UnitedLinux 1.0 - Solaris 7 - Debian Woody (patched to half Sarge ;-)) the cool thing is that we had very little trouble to get 1.38 packages going in that OS zoo. Only two machines required a self compiled static FD: the UnitedLinux one and SuSE 6.4 (which is ok, since those are way too old). The debians are not upgraded yet. Many thanks to the guys who make all the packages. This is awesome work! MY SOLUTION Make the fileset with 'onfs=no' and include '/'. Then use regular expressions to exclude everything but the things I want. Here is what I have so far: FileSet { Name = "Data-UNIX-FileServer" Ignore FileSet Changes = yes# can't effort a full right now Include { Options { signature=MD5 onefs = no # cross all filesystems fstype = ext2 # should avoid possible NFS mounts? You might want to add other FSes as well, depending on the possibility that you also have to back up BSD or Mac or Sun or whatever systems. yeah, good point. I forgot about the solaris ones. [...] this seems to work so far. I tested it with estimate listing on 1.38.0 (FD, Dir, SD) and the list looks very promising. I did not actually run a backup job yet. Well, I didn't try your fileset, but it does look promising. It seems that you understood the two options blocks - default and more detailed ones - correctly which I never did :-) aehmmm... well, now as you say that. I wasn't aware of any 'block ranking'/default or detailed blocks etc . ;-) Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] FileSet accross mountpoints with regular expressions
Hello, I do not have a problem in the sense that something does not work - I just would like to hear some opinions as I am not sure how fool proof my solution is: SITUATION There are multiple clients and I like to have one fileset for all of them. As they are quite similar. Each one looks like /u01 /u02 ... /u09 /u10 ... /uNN the number of /uNN directories is variable. But not more than 2 digits. Each /uNN directory is a mountpoint. It is never NFS. Always local disks. I want to avoid the File = "\\|sh -c \"" " approach as it would IMHO cause me a headache to write a shell command that works on all my OSes with different shell/find/awk/... versions. MY SOLUTION Make the fileset with 'onfs=no' and include '/'. Then use regular expressions to exclude everything but the things I want. Here is what I have so far: FileSet { Name = "Data-UNIX-FileServer" Ignore FileSet Changes = yes# can't effort a full right now Include { Options { signature=MD5 onefs = no # cross all filesystems fstype = ext2 # should avoid possible NFS mounts? regex = "^\/u[0-9]{2}" # /u01 to /uNN directories } Options { # exclude just everything exclude = yes regex = ".*" } # the regular expressions will filter things File = / } } this seems to work so far. I tested it with estimate listing on 1.38.0 (FD, Dir, SD) and the list looks very promising. I did not actually run a backup job yet. can anyone see any problems with this procedure? Maybe beside possible mount-loops because of onefs=yes. I tried to limit that possibility further down by adding fstype=ext2. Is that supposed to work as I expect? a side note: when adding 'Include=yes' to the first options resource bacula gives me a syntax error saying that this is not known at this position. According to one sample in the manual it /should/ be valid? However I can not find the include option mentioned in the options list - so I would guess that the FileSet example is outdated. best regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Stephan Ebelt wrote: Another test would be to try the same with RunAfterJob. And see what happens then. I'll do that later. ok, just tried. It behaves similar with one difference: the RunAfterJob failure is reported at the end of the output. Not at the beginning as with ClientRunAfterJob. 22-Sep 16:06 dd-lx-backup.dd.net-linx: RunBefore: run before job: exit with 0 22-Sep 16:06 dd-lx-backup.dd.net-linx: Start Backup JobId 1049, Job=testjob.2005-09-22_16.06.39 22-Sep 16:06 dd-lx-backup.dd.net-linx: Volume "disk4-0009" previously written, moving to end of data. 22-Sep 16:06 dd-lx-backup.dd.net-linx: Bacula 1.36.3 (22Apr05): 22-Sep-2005 16:06:44 JobId: 1049 Job:testjob.2005-09-22_16.06.39 Backup Level: Incremental, since=2005-09-22 16:04:18 Client: dd-lx-backup FileSet:"SystemConfig-UNIX" 2005-08-05 21:36:15 Pool: "disk4" Storage:"disk4" Start time: 22-Sep-2005 16:06:40 End time: 22-Sep-2005 16:06:44 FD Files Written: 1 SD Files Written: 1 FD Bytes Written: 15,427,565 SD Bytes Written: 15,427,667 Rate: 3856.9 KB/s Software Compression: None Volume name(s): disk4-0009 Volume Session Id: 57 Volume Session Time:1127203794 Last Volume Bytes: 39,266,431,731 Non-fatal FD errors:0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination:Backup OK 22-Sep 16:06 dd-lx-backup.dd.net-linx: Begin pruning Jobs. 22-Sep 16:06 dd-lx-backup.dd.net-linx: No Jobs found to prune. 22-Sep 16:06 dd-lx-backup.dd.net-linx: Begin pruning Files. 22-Sep 16:06 dd-lx-backup.dd.net-linx: No Files found to prune. 22-Sep 16:06 dd-lx-backup.dd.net-linx: End auto prune. 22-Sep 16:06 dd-lx-backup.dd.net-linx: RunAfter: run after job: exit with 5 22-Sep 16:06 dd-lx-backup.dd.net-linx: testjob.2005-09-22_16.06.39 Warning: RunAfterJob error: ERR=Child exited with code 5 so the return code is recognized. But the job does not fail as described in the manual. Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Martin Simmons wrote: Stephan> that means it always terminates with '*** Backup Error ***' when you Stephan> exit with e.g. 5? Correct -- a non-zero exit code from the ClientRunBeforeJob always results in '*** Backup Error ***'. Also, I always get 'Backup OK' regardless of the exit code from ClientRunAfterJob. exactly the same here. [...] Stephan> Or do you mean that the return code of ClientRunBeforeJob is send to the Stephan> director before the ClientRunAfterJob is executed? That would make sense. No, I meant the count of "Non-fatal FD errors" (which changes 'Backup OK' to 'Backup OK -- with warnings' when > 0). You said that this count is 1 when your ClientRunAfterJob script returns certain error codes, but I don't see how that can happen. ah, I see. I tried again: one job with ClientRunBeforeJob returning 0 and ClientRunAfterJob returning 5. The result was 'Backup OK'. So there must have been something else in the previous try. That's my fault then. Here's the output: 22-Sep 15:11 dd-lx-backup.dd.net-linx: Start Backup JobId 1047, Job=DD31.2005-09-22_15.11.28 22-Sep 15:11 dd-lx-backup.dd.net-linx: Volume "disk4-0009" previously written, moving to end of data. 22-Sep 15:11 dd-lx-oracle3.dd.net-linx: DD31.2005-09-22_15.11.28 Fatal error: ClientRunAfterJob returned non-zero status=268435461. ERR=Child exited with code 5 22-Sep 15:11 dd-lx-backup.dd.net-linx: Bacula 1.36.3 (22Apr05): 22-Sep-2005 15:11:31 JobId: 1047 Job:DD31.2005-09-22_15.11.28 Backup Level: Incremental, since=2005-09-21 10:53:10 Client: dd-lx-oracle3 FileSet:"Data-UNIX-Oracle" 2005-08-05 19:00:01 Pool: "disk4" Storage:"disk4" Start time: 22-Sep-2005 15:11:29 End time: 22-Sep-2005 15:11:31 FD Files Written: 0 SD Files Written: 0 FD Bytes Written: 0 SD Bytes Written: 0 Rate: 0.0 KB/s Software Compression: None Volume name(s): disk4-0009 Volume Session Id: 54 Volume Session Time:1127203794 Last Volume Bytes: 39,178,621,354 Non-fatal FD errors:0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination:Backup OK [...] Stephan> hmmm, from how I understood the manual it should work exactly like this? Stephan> It says: Stephan> Run After Job = Stephan> The specified command is run as an external program after the Stephan> current job terminates. This directive is not required. The command Stephan> string must be a valid program name or name of a shell script. If Stephan> the exit code of the program run is non-zero, the current Bacula job Stephan> will terminate in error. Before submitting the specified command to Stephan> the operating system, Bacula performs character substitution as Stephan> described above for the Run Before Job directive. Stephan> An example of the use of this command is given in the Tips Chapter Stephan> of this manual. As of version 1.30, Bacula checks the exit status of Stephan> the RunAfter program. If it is non-zero, the job will be terminated Stephan> in error. Stephan> [...] Stephan> Client Run After Job = Stephan> This command is the same as Run After Job except that it is run on Stephan> the client machine. Note, please see the notes above in Client Run Stephan> Before Job concerning Windows clients. I've never seen it behave like that, so I suspect the doc is wrong. hmm, before we correct the doc I guess it would be worth thinking about modifing the code since the doc appears to make a lot of sense? To me in that situation at least. Should I log this as a bug? Another test would be to try the same with RunAfterJob. And see what happens then. I'll do that later. Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Martin Simmons wrote: >> Interesting. What is the largest number that behaves like this? Stephan> looks like it's 199. I suppose that makes sense in a warped kind of way :-) ;-) >> Does returning 5 like you did initially always return in "Backup OK -- with >> warnings" with "Non-fatal FD errors: 1"? Stephan> as ClientRunAfterJob it does. Ah, sorry, I misread your original post and was looking at ClientRunBeforeJob. However, I still don't get any Non-fatal errors on FreeBSD with the ClientRunAfterJob. that means it always terminates with '*** Backup Error ***' when you exit with e.g. 5? In fact, it should be impossible because that number is returned to the Director before the ClientRunAfterJob runs! which number? Sorry, I may misunderstood something: but the Director should not know the return code /before/ the ClientRunAfterJob runs? (that would be an impressive piece of code - predicting the future ;-) ) Or do you mean that the return code of ClientRunBeforeJob is send to the director before the ClientRunAfterJob is executed? That would make sense. Stephan> hmmm, in my case not. Since I may have stored something that is not Stephan> (entirely) useful. Stephan> The background of this is that my ClientRunBeforeJob initiates a Oracle Stephan> database full export. The actual data is written to a FIFO pipe where Stephan> bacula sucks the data from during the backup job. This works pretty well Stephan> so far. Stephan> Finally I use the ClientRunAfterJob to parse the log file that oracle Stephan> creates. Since sometimes the database is not /entirely/ exported. The Stephan> dump might not contain everything we would need/want to restore. E.g. Stephan> somebody could have locked a single table. Or oracle produced a variety Stephan> of other warnings. In that case it is necessary to run the export again. Stephan> Thus the backup job should fail and notify us this way. Ah, I see. I don't know of any way to make that situation record a failed job. hmmm, from how I understood the manual it should work exactly like this? It says: Run After Job = The specified command is run as an external program after the current job terminates. This directive is not required. The command string must be a valid program name or name of a shell script. If the exit code of the program run is non-zero, the current Bacula job will terminate in error. Before submitting the specified command to the operating system, Bacula performs character substitution as described above for the Run Before Job directive. An example of the use of this command is given in the Tips Chapter of this manual. As of version 1.30, Bacula checks the exit status of the RunAfter program. If it is non-zero, the job will be terminated in error. [...] Client Run After Job = This command is the same as Run After Job except that it is run on the client machine. Note, please see the notes above in Client Run Before Job concerning Windows clients. Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Martin Simmons wrote: >> Well I can't repeat this here with 1.36.3 on FreeBSD. What should happen is >> that the director says >> >> Fatal error: FD gave bad response to ClientRunBeforeJob command: wanted 2000 OK RunBefore >> got: 2905 Bad RunBeforeJob command. Stephan> now we get closer: I modified my ClientRunBeforeJob to exit with 100. Stephan> And I get exactly what your director says. Here's the output: Interesting. What is the largest number that behaves like this? looks like it's 199. But I did not try everything in between. ;-) 200 gives: 21-Sep 16:36 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.36.43 Fatal error: ClientRunBeforeJob returned non-zero status=268435656. ERR=Permission denied 201 gives: 21-Sep 16:39 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.39.19 Fatal error: ClientRunBeforeJob returned non-zero status=268435657. ERR=Exec format error 202 gives: 21-Sep 16:40 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.39.58 Fatal error: ClientRunBeforeJob returned non-zero status=268435658. ERR=Bad address and 205 is: 21-Sep 16:46 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.46.30 Fatal error: ClientRunBeforeJob returned non-zero status=268435661. ERR=File name too long ... Does returning 5 like you did initially always return in "Backup OK -- with warnings" with "Non-fatal FD errors: 1"? as ClientRunAfterJob it does. as ClientRunBeforeJob returning 5 behaves as expected - the Job fails with fatal error. The message in that case is rather normal: 21-Sep 16:48 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_16.48.18 Fatal error: ClientRunBeforeJob returned non-zero status=268435461. ERR=Child exited with code 5 Stephan> when I do the same thing as ClientRunAfterJob it does /not/ error out Stephan> like this [1]. So it looks like there's a difference between Stephan> ClientRunBeforeJob and ClientRunAfterJob return codes. Yes, the result of ClientRunAfterJob is probably ignored because it is pointless making the job fail after it has finished. hmmm, in my case not. Since I may have stored something that is not (entirely) useful. The background of this is that my ClientRunBeforeJob initiates a Oracle database full export. The actual data is written to a FIFO pipe where bacula sucks the data from during the backup job. This works pretty well so far. Finally I use the ClientRunAfterJob to parse the log file that oracle creates. Since sometimes the database is not /entirely/ exported. The dump might not contain everything we would need/want to restore. E.g. somebody could have locked a single table. Or oracle produced a variety of other warnings. In that case it is necessary to run the export again. Thus the backup job should fail and notify us this way. Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Martin Simmons wrote: >> Which version of bacula is running on dd-lx-oracle3? Stephan> sorry, I totaly forgot: it's the bacula-client-1.36.3-1 RPM package Stephan> running on FC3. Well I can't repeat this here with 1.36.3 on FreeBSD. What should happen is that the director says Fatal error: FD gave bad response to ClientRunBeforeJob command: wanted 2000 OK RunBefore got: 2905 Bad RunBeforeJob command. now we get closer: I modified my ClientRunBeforeJob to exit with 100. And I get exactly what your director says. Here's the output: 21-Sep 14:54 dd-lx-backup.dd.net-linx: Start Backup JobId 1015, Job=DD31.2005-09-21_14.54.35 21-Sep 14:54 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_14.54.35 Fatal error: ClientRunBeforeJob returned non-zero status=268435556. ERR=Child exited with code 100 21-Sep 14:54 dd-lx-backup.dd.net-linx: DD31.2005-09-21_14.54.35 Fatal error: FD gave bad response to ClientRunBeforeJob command: wanted 2000 OK RunBefore got: 2905 Bad RunBeforeJob command. 21-Sep 14:54 dd-lx-backup.dd.net-linx: DD31.2005-09-21_14.54.35 Error: Bacula 1.36.3 (22Apr05): 21-Sep-2005 14:54:39 JobId: 1015 Job:DD31.2005-09-21_14.54.35 Backup Level: Incremental, since=2005-09-21 10:53:10 Client: dd-lx-oracle3 FileSet:"Data-UNIX-Oracle" 2005-08-05 19:00:01 Pool: "disk4" Storage:"disk4" Start time: 21-Sep-2005 14:54:37 End time: 21-Sep-2005 14:54:39 FD Files Written: 0 SD Files Written: 0 FD Bytes Written: 0 SD Bytes Written: 0 Rate: 0.0 KB/s Software Compression: None Volume name(s): Volume Session Id: 33 Volume Session Time:1127203794 Last Volume Bytes: 0 Non-fatal FD errors:0 SD Errors: 0 FD termination status: SD termination status: OK Termination:*** Backup Error *** when I do the same thing as ClientRunAfterJob it does /not/ error out like this [1]. So it looks like there's a difference between ClientRunBeforeJob and ClientRunAfterJob return codes. Stephan [1] thats what I need since my ClientRunAfterJob parses a log file that is created during ClientRunBeforeJob - only then I know whether that backup has failed --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] ClientRunAfterJob return code
Stephan> when returning '-1' the job ends like this (Backup OK): Exit status values should be in the range 0 to 255. What happens if you use exit 255? 'exit 255' looks similar to 'exit -1': 21-Sep 10:41 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_10.41.28 Fatal error: ClientRunAfterJob returned non-zero status=268435711. ERR=Unknown error during program execvp interesting might be that an 'exit 100' looks like: 21-Sep 10:48 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_10.48.09 Fatal error: ClientRunAfterJob returned non-zero status=268435556. ERR=Child exited with code 100 while an 'exit 200' ends up with: 21-Sep 10:49 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_10.49.23 Fatal error: ClientRunAfterJob returned non-zero status=268435656. ERR=Permission denied is bacula somehow intepreting the codes? The other difference between your two jobs is that one has been upgraded to Full. Maybe that causes the different behaviour? that should not be? shouldn't it? I monitor that problem for quite a while now - and I can't see a difference between Full/Incremental/Differential. Did you get some other warning in the second job? nope. I just removed the volume recycling/pruning messages from the job report. Everything else is there. Here's the complete output from the exit 255 test: 21-Sep 10:47 dd-lx-backup.dd.net-linx: Start Backup JobId 1007, Job=DD31.2005-09-21_10.47.28 21-Sep 10:47 dd-lx-backup.dd.net-linx: Volume "disk4-0008" previously written, moving to end of data. 21-Sep 10:47 dd-lx-oracle3.dd.net-linx: DD31.2005-09-21_10.47.28 Fatal error: ClientRunAfterJob returned non-zero status=268435710. ERR=Unknown error during program execvp 21-Sep 10:47 dd-lx-backup.dd.net-linx: Bacula 1.36.3 (22Apr05): 21-Sep-2005 10:47:32 JobId: 1007 Job:DD31.2005-09-21_10.47.28 Backup Level: Incremental, since=2005-09-21 10:41:30 Client: dd-lx-oracle3 FileSet:"Data-UNIX-Oracle" 2005-08-05 19:00:01 Pool: "disk4" Storage:"disk4" Start time: 21-Sep-2005 10:47:30 End time: 21-Sep-2005 10:47:32 FD Files Written: 0 SD Files Written: 0 FD Bytes Written: 0 SD Bytes Written: 0 Rate: 0.0 KB/s Software Compression: None Volume name(s): disk4-0008 Volume Session Id: 25 Volume Session Time:1127203794 Last Volume Bytes: 15,614,658,447 Non-fatal FD errors:0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination:Backup OK 21-Sep 10:47 dd-lx-backup.dd.net-linx: Begin pruning Jobs. 21-Sep 10:47 dd-lx-backup.dd.net-linx: No Jobs found to prune. 21-Sep 10:47 dd-lx-backup.dd.net-linx: Begin pruning Files. 21-Sep 10:47 dd-lx-backup.dd.net-linx: No Files found to prune. 21-Sep 10:47 dd-lx-backup.dd.net-linx: End auto prune. Which version of bacula is running on dd-lx-oracle3? sorry, I totaly forgot: it's the bacula-client-1.36.3-1 RPM package running on FC3. Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] ClientRunAfterJob return code
Hello, maybe this issue has been clarified already - I just couldn't find it. Please don't spend time reading further if that is the case. Just give me a hint where to find the discussion. My understanding from the manual is that the entire Job will (should?) terminate as '*** Backup Error ***' if the ClientRunAfterJob command returns something non-zero. I am using a bash script as ClientRunAfterJob and tried two things: exit -1 and exit 5 when returning '-1' the job ends like this (Backup OK): [...] 15-Sep 21:05 dd-lx-oracle3.dd.net-linx: ClientRunAfterJob: Thu 15 Sep 2005 09:05:46 PM CEST - start backup check of Instance: DD31 15-Sep 21:05 dd-lx-oracle3.dd.net-linx: ClientRunAfterJob: Database Export of database DD31 was not successfull! Please check the logfile... 15-Sep 21:05 dd-lx-oracle3.dd.net-linx: DD31.2005-09-15_19.00.00 Fatal error: ClientRunAfterJob returned non-zero status=268435711. ERR=Unknown error during program execvp 15-Sep 21:05 dd-lx-backup.dd.net-linx: Bacula 1.36.3 (22Apr05): 15-Sep-2005 21:05:46 JobId: 876 Job:DD31.2005-09-15_19.00.00 Backup Level: Full (upgraded from Incremental) Client: dd-lx-oracle3 FileSet:"Data-UNIX-Oracle" 2005-08-05 19:00:01 Pool: "disk4" Storage:"disk4" Start time: 15-Sep-2005 19:00:01 End time: 15-Sep-2005 21:05:46 FD Files Written: 1 SD Files Written: 1 FD Bytes Written: 5,558,467,555 SD Bytes Written: 5,558,467,641 Rate: 736.7 KB/s Software Compression: 81.1 % Volume name(s): disk4-0004 Volume Session Id: 292 Volume Session Time:1125570766 Last Volume Bytes: 43,821,823,021 Non-fatal FD errors:0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination:Backup OK [...] when returning '5' it ends like this (Backup OK - with warnings) [...] 20-Sep 10:11 dd-lx-oracle3.dd.net-linx: DD31.2005-09-20_10.08.53 Fatal error: ClientRunAfterJob returned non-zero status=268435461. ERR=Child exited with code 5 20-Sep 10:11 dd-lx-backup.dd.net-linx: Bacula 1.36.3 (22Apr05): 20-Sep-2005 10:11:10 JobId: 983 Job:DD31.2005-09-20_10.08.53 Backup Level: Incremental, since=2005-09-15 19:00:01 Client: dd-lx-oracle3 FileSet:"Data-UNIX-Oracle" 2005-08-05 19:00:01 Pool: "disk4" Storage:"disk4" Start time: 20-Sep-2005 10:09:55 End time: 20-Sep-2005 10:11:10 FD Files Written: 0 SD Files Written: 0 FD Bytes Written: 0 SD Bytes Written: 0 Rate: 0.0 KB/s Software Compression: None Volume name(s): disk4-0008 Volume Session Id: 1 Volume Session Time:1127203794 Last Volume Bytes: 385 Non-fatal FD errors:1 SD Errors: 0 FD termination status: OK SD termination status: OK Termination:Backup OK -- with warnings [...] Bacula is clearly detecting the return code. But the Job Termination is always 'Backup OK' (with or without warnings). I would expect to see '*** Backup Error ***' in that case. Is that something that should not be? Or is just my expectation wrong? (please note that the second output was a test where I modified the script to exit right away - so it did not produce any output and no data was transfered) thanks and regards, Stephan --- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] FIFO permissions during restore
Hello, we are using bacula to dump our oracle databases. This works very well with the FIFO feature and a shell tool as RunClientBeforeJob. What I want now is that our DBAs (who use the 'oracle' account, they do not have root access) are able to restore their backups independend from technical staff. This works very nice with a restricted console. There is just a little permission problem in the way. I believe. I'll try to describe it below. oh one more thing: the bacula director and sd version is 1.36.2 on FC3. The file daemon version is 1.36.1 on FC1 x86_64. This is how the fifo looks like in the bacula restore command. The permissions are shown as they should be. $ cd tmp cwd is: /tmp/ $ dir frw-rw-r-- 1 root bacula 0 2005-03-30 02:54:19 /tmp/oracle-backup.pipe when I now mark this file and start the restore job the pipe will appear on the target system like this: # ls -l /tmp/*pipe prw-r- 1 root bacula0 Apr 1 16:38 oracle-backup.pipe I now assume that bacula will set the permissions right /after/ the file content has been restored. And that seems to be the problem in my situation. Because the shell tool (to suck the data from the pipe and import it straight into oracle) will run as user 'oracle' which is treated as 'other' in that case. So the tool fails with permission denied. Well, I could make oracle member of the bacula group. But that would permit access to other things as well? Also they run into the problem when restoring the dump on another server. Since group IDs are not identical... Does anyone have an idea on this? thanks in advance, Stephan --- This SF.net email is sponsored by Demarc: A global provider of Threat Management Solutions. Download our HomeAdmin security software for free today! http://www.demarc.com/Info/Sentarus/hamr30 ___ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users