From: Kern S. <ke...@si...> - 2005-02-27 21:37:57
|
On Sunday 27 February 2005 19:26, Arno Lehmann wrote: > Hi, > > Thomas Mangin wrote: > >> Use purge, prune or delete commands. > > > > I was just "put off" using the purge command due to the numerous "are > > you sure of what you are doing" warnings in the doc, but I guess I will > > have to ignore them and use it even if I am not sure of what I am doing > > ..... I will spend some more time reading the manual - which is not the > > clearest on this part. > > Well, I'm not using file storage, so I can't give you any definitive > answers. But... > > First, I suggest you simply try it :-) > Afterwards, you know. And then submit your findings as documentation > improvements! > > > As I am using a pool with different file for full/inc/diff will the > > space freed will be re-used ? > > I'm not sure I understand what you mean. > > I see two possibilities: > Either a complete volume file is pruned/purged. Then that whole file can > be reused (or deleted). > Or you free some space inside one volume. In this case, I guess bacula > would not reuse it - after all, volume files are /only/ a replacement > for tapes, and you don't use tapes that way. > > > Will the purge command cause the created volume on the HD to be deleted > > /set to 0 bytes ? > > Don't know - try it. No. > > > As well, I am wondering what happen if when a file exist and is re-used > > should the next backup be shorter. Will the free space be left on the > > file used or will the file be truncated ? > > I guess the file will be truncated when reused and thus be shorter. But > - try it. Yes, it is truncated when it is "recycled". > > >>> 2 - What could I do to reduce the space usage on the HD of current > >>> backup where I changed the fileset to be smaller ? > >> > >> Use compression. Use smaller volumes, so that pruned jobs lead to > >> faster volume recycling. > > > > I use compression to transfer file - defined in the fileset, so I would > > expect to see it on the volume. Is it what you are telling me, I did not > > see any "compression" option at SD/volume level. > > No, compression is demanded in the filesets. With bacula, it's always > the client doing the compression, so the SD only has to work with data > streams and needn't know anything about their contents. > > >>> 4 - What would be the correct command to tell the console to re-load > >>> the configuration file once the current backup are finished ? > >> > >> reload > > > > Ok, I was thinking that it may have been possible through the console ;) > > something like : > > > > bin/bconsole -c bin/bconsole.conf <<END_OF_DATA > > @output /dev/null > > messages > > @output /tmp/log1.out > > wait > > reload messages > > @output > > quit > > END_OF_DATA > > > > Oups, Look like a feature request ;) > > Ahm. No. > I think my answer was too short. > Once again: > either, enter "reload" at any console command prompt. Or give the > director a HUP signal (I think that works...) > > To modify your example: > echo "reload"|bconsole -c /etc/bacula/bconsole.conf Yes. > > If the reload can't be done immediately because the resources are in > use, it will be queued automatically, so no need for "wait". Correct. The reload is "smart" any running jobs continue with the old config. Any new jobs run with the new config. Any change to the schedules *may* not take place for 1 hour since jobs are pre-scheduled up to 1 hour in advance. No one has noticed this and complained, so I haven't fixed it. :-) > > Very important: Verify the configuration's correctness before reloading, > the director crashes when confronted with a configuration error. The -t > command switch is your friend... > > > The manual is pretty good and complete, I was quite impressed by it, but > > sometime it is so complete that you do miss information as it is not > > where you expect it and sometime it expect you to know more than what > > you do. But overall it is as impressive as bacula itself. The new manual (1.37) will be much better because it has an index (still being constructed). This should help a lot with a manual as big as Bacula's. > > Too true. > > Well, my experience is that you need some time for experiments. After an > initial reading of the manual and a test setup and some observation what > actually happens the understanding grows. Then, the manual is a good > reference :-) Yup :-) > > > Arno > > > Regards, > > > > Thomas |