Dustin J. Mitchell schrieb: > On Thu, Oct 1, 2009 at 11:44 AM, Sebastian Henrich <s...@gmx.de> wrote: > >> 1. If I call amcleanup or amstatus the following error is returned: >> >> Can't exec "-eo": Datei oder Verzeichnis nicht gefunden at >> /usr/lib/perl5/vendor_perl/5.8.8/Amanda/Process.pm line 177. >> -eo pid,ppid,command: Datei oder Verzeichnis nicht gefunden at >> /usr/lib/perl5/vendor_perl/5.8.8/Amanda/Process.pm line 177. >> > > This is due to problems invoking "ps", to try to get a list of active > Amanda processes. Can you take a look at Amanda::Process, and at your > own configuration (is $Amanda::Constants::PS empty??), and the > behavior of your local "ps" executable, and see if you can suggest a > fix? > As I can see /usr/lib/perl5/vendor_perl/5.8.8/Amanda/Process.pm imports the Constants from /usr/lib/perl5/vendor_perl/5.8.8/Amanda/Constants.pm. Line 83 of Constants.pm looks like $PS = "". After replacing it with $PS = "ps" amcleanup and amstatus are working as expected. If this can by caused during building the rpm I can try to contact the package maintainer. >> 2. The level 1 backups are as big as the level 0 backups: >> > > Hmm, what are you using to do backups? Did you look in some of the > debug logs to see if you can find out why? > > I'm using gnutar for my backups. Inside sendsize.20091001175625.debug (Second backup) I found the following line
gnutar: error opening /var/lib/amanda/gnutar-lists/servlnx01.sebhen.lan_home_0: No such file or directory I think this is caused by setting "record no" inside dumptype global. Is this the cause why the incremental backups are as big as the full dumps? I'll reset my configuration and try again. >> 3. If I choose request="tty_email" in changer.conf the changer skript >> sent me e-mails requesting for a new tape during the second backup but >> aborted after lastslot times. If I choose request="email" everything >> works fine. The second backup was fired up by cron, the first was >> manually initiated. >> > > Hmm, I'm not sure I understand here -- where is there a lastslot > configuration for chg-manual? What exactly was the problem, if it > sent you emails for each new tape? > > My changer.conf looks like this (I think I found this inside the wiki): ----8<-----8<----- resend_mail=900 # 15 minutes timeout_mail=604800 # 7 days request=email # Send an email to ask the user to change tape. firstslot=1 lastslot=8 ----8<-----8<----- I set lastslot to 8 because I only use 8 tapes. If I don't set lastslot and call "amtape test show", amtape iterates through 99 slots because lastslot is set to 99 by default. When no space is left on the fist device I get an email with the following content: ----8<-----8<----- Insert Amanda tape into slot 2 (tape:/dev/nst0l) or `touch /etc/amanda/test/chg-manual.abort` to abort. ----8<-----8<----- 8 minutes later I get another e-mail looking like this ----8<-----8<----- Can't open tape device /dev/nst0l: No medium found Insert Amanda tape into slot 3 (tape:/dev/nst0l) or `touch /etc/amanda/test/chg-manual.abort` to abort. ----8<-----8<----- Every 8 minutes later I get an new mail until slot 8 is reached. After another 8 minutes I get this e-mail ----8<-----8<----- Hostname: servlnx01 Org : servlnx01 Config : test Date : October 1, 2009 These dumps were to tape Test-07. *** A TAPE ERROR OCCURRED: [No more writable valid tape found]. There are 24869146k of dumps left in the holding disk. Run amflush to flush them to tape. The next 2 tapes Amanda expects to use are: 1 new tape, Test-01. The next new tape already labelled is: Test-08. FAILURE DUMP SUMMARY: servlnx01.sebhen.lan /home lev 1: partial taper: No space left on device ... ----8<-----8<----- I'm wondering about the 8 minutes I expected 15 minutes between the mails because of "resend_mail=900". I also expected that I'll get an e-mail with aborting information after 7 days an not after 64 minutes. I took a look at the changer scripts which underlines my expectations. If I'm fast enough and replace the tape before I get the last e-mail the backups completes correct. So I'm confused at this point. Thanks for your help Sebastian
