Re: [Safekeep-devel] Rdiff-backup stuck during backup

[Frank C. <fr...@cr...>]

Jan,

This is more an issue with rdiff-backup than safekeep and may be better
discussed on the rdiff-backup list (<rdi...@no...>).

Two things to do would be also bump up the verbosity on safekeep (i.e.
multiple -v's on the run) as well for rdiff-backup, which you have
done, and to send us more of the output to see if we can see anything
else.

Also, don't discount an OS bug, as I remember seeing reports of issues
with futex kernel locking recently, which could cause issues like this.

Given it is different files each time, it is probably not data
dependent, but may be either time or a race condition.

Sorry, can't help more at the moment, as there aren't enough details.

Regards
Frank

On Fri, 2015-08-14 at 16:33 +0200, Jan Škoda wrote:
> Hello,
> 
> I am using safekeep for filesystem backup through rdiff-backup. The 
> backup process however often hangs in the middle of a file (different 
> file every time) for some servers. With '-v6' option log contains:
> 
>  > ...
> > Processing changed file srv/mapcache/cz/bku_orto1.db
> > Regular copying ('srv', 'mapcache', 'cz', 'bku_orto1.db') 
> > to/disk0/backup/cache2.geo--cache/srv/mapcache/cz/rdiff
> > -backup.tmp.5
> 
> The file:
> > -rw-rw-r-- 1 boris mapcache 4.5G Dec  2  2014 
> > /srv/mapcache/cz/bku_orto1.db
> Resulting tmp file:
> > -rw------- 1 safekeep nogroup 1.4G Aug 14 15:31 rdiff-backup.tmp.5
> 
> According to strace, all processes (safekeep and rdiff-backup) are 
> waiting for input through some pipe. I use trickle and almost default 
> configuration of safekeep (just backup.user and base.dir set).
> 
> Does anyone have an idea how to debug or fix this problem?
> 
> Thank you!
> --
> Jan Škoda
> 
> 
> ---------------------------------------------------------------------
> ---------
> _______________________________________________
> Safekeep-devel mailing list
> Saf...@li...
> https://lists.sourceforge.net/lists/listinfo/safekeep-devel