On Thursday 23 June 2005 02:22, Ross Boylan wrote: > I restarted the sd demon, and it solved the problem. A few questions > appear below. > > On Wed, Jun 22, 2005 at 09:41:38PM +0200, Arno Lehmann wrote: > > Hi, > > > > Ross Boylan wrote: > > >As detailed in my earlier messages, I have my storage writing to disk > > >files, and my first attempt failed because I had an invalid volume > > >name. > > > > > >I have cancelled the offending jobs (except parts linger on--see > > >later), fixed (maybe) my label generation, and asked the director to > > >reload its configuration (I'm not sure if that's a full restart under > > >the hood or not--this is on Debian). > > > > One important question: Is anything vital running in bacula at the > > moment? The most effective way to to get everything unstuck is, after > > all, a /etc/init.d/bacula restart or killall -9 bacula :-) > > No. I'm still at the experimental phase. > > Speaking of which, when I finally stop mucking around, what's the best > way to cleanup. I see "delete" and "purge" commands. It's hard to > tell what the difference is. Perhaps delete on a volume just deletes > the volume, while purge clears the related file and job entries too? > I take it I also need to delete the actual files (the "volumes" on my > disk) to get rid of them. > > I believe the standard advice is to rerun the database creation > scripts, but they are tied into the Debian package installation > procedures, so I'm reluctant to go that route. > > > On the clients with stuck fd too, of course... > > > > Usually, a reload (or reload from inside a console) is not a restart > > but > > This is the second answer that refers to reload from the console, but > I don't see any such command in the manual. Am I missing something? > > What I meant was a did the Debian-specific > > /etc/init.d/bacula-dir force-reload > > which I assume is sending a signal to the process. > > To restart the sd I did > /etc/init.d/bacula-sd restart > > > only triggers reloading of the configuration (this works in the director > > only!).
> > Usually, the reload will be queued, because (my personal guess > > without any real knowledge about the source) it might change important > > data in running or queued jobs. This is true, there is a queuing process if you want to call it that. However, the reload does take effect immediately. What is queued is cleanup and releasing of the old config information. The Director waits for the last running job to exit before the old conf info is released. You can do at most 10 such queuing of old config data before the Dir says no to a reload request, which is *really* unlikely in a real situation unless you are reloading a lot while a long job is running. > > Hadn't thought of that.... > > I also hope that randomly restarting particular demons won't screw up > the other ones. You can kill off individual daemons at any time without screwing up the other ones. In some cases, depending on the way the OS handles the networking, it may take a couple of hours before the other daemons "cleanup" any jobs that were running. > ... > > > >I do not want to use label from the console, because it is the > > >automatic labelling that I need to test (I also don't know if that > > >would help things). > > > > Understandable. Still, why not simply shutdown bacula and restart it? > > I did. > > Out of curiosity, does anyone know if doing a manual label would have > got the storage demon going? > > > The stuck jobs on remote machines will usually time out after 2 hours. > > If they don't it should be safe to restart their daemons. > > All on one machine before I try anything fancy :) > > > Concerning a stuck SD - that's tricky. I managed that myself some times, > > but it always involved tape devices here. In those cases, there was > > either a really long timeout necessary (long in computer terms - we're > > talking about some hours here) or I turned the drive's power off, waited > > for the kernel to start complaining, turned it on again. After that, the > > SD recovered quickly. > > Even though I tried some of the tape-oriented commands (unmount, > release) this was just a wild guess. I don't know what, if anything, > they do with file backup. Anybody know? > > > > ------------------------------------------------------- > SF.Net email is sponsored by: Discover Easy Linux Migration Strategies > from IBM. Find simple to follow Roadmaps, straightforward articles, > informative Webcasts and more! Get everything you need to get up to > speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users -- Best regards, Kern ("> /\ V_V ------------------------------------------------------- SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
