[Dirvish] Weird problems
dirvish at hunsperger.com
Mon May 8 07:37:10 UTC 2006
On Mon, May 08, 2006 at 08:09:44AM +0200, Matthew Whittaker-Williams wrote:
> Good morning ( CEST ),
> I`ve got this weird problem with running dirvish by cron.
> It runs most of the time just fine, but the last couple of months the
> actual backup seems to be done 2 hours later than the dirvish-runall
> command is run.
> Have any of you seen and experienced this behaviour of dirvish before?.
If you are saying that the backup finishes 2 hours after dirvish-runall
starts, yes, this can be quite normal. My system takes about 3 hours
to do a full backup, assuming not much has changed.
> So it seems that dirvish-expire is taking quiet some time to finish
> before the actual backup is run.
> Is this behaviour normal with like +- 50 machine`s to backup?
It is really just an issue of how many files you have, and how quickly
your file system can decrement hardlink counts and remove files. On the
backup server I run, I have seen it take over an hour to remove a single
vault image, but I'm backing up 3000ish user accounts.
> Cron did run but no backup has been made *shrug*, this actually happend
> before but i didn`t managed to replicate this behaviour.
> Now I didn`t get any output either from the dirvish-runall command so I
> am kind of stuck where to look at this problem.
> When I looked at the running processes I haven`t seen any dirvish
> commands come by so I think its broken some where.
> So the question here is again have any of you seen this behaviour before?
Your run script will only run dirvish-runall if dirvish-expire finished
without errors. Is this happening? Also, I presume you used ps to look
for dirvish processes, and not top?
More information about the Dirvish