On 09/23/10 14:00, Sean Clark wrote:
> I've been setting maximum jobs to 2 for most file daemons around here,
> since we should never actually be running more than 1 job (but we have
> room to run a second simultaneous job in case I ever think of a reason
> that we need to).
>
> Having mistakenly forgotten to set "Allow Duplicate Jobs=no", I now have
> a user whose laptop is running two simultaneous "Full" 100GB+ backups.
> I went to cancel the later-starting of the two and...I'm denied. The
> director appears to hang for several minutes before finally giving me a
> "Like, Dude, Something Went Wrong!" message. The error message suggests
> the problem is that I've hit the "maximum concurrent jobs" limit on the
> FD (as I'm pretty confident that of the other two options, the
> passwords match and the two simultaneously-running jobs suggest there's
> nothing currently wrong with the networking...).
>
> I find I can't even get client status. "status client=(hostname)" for
> that system gives me:
>
> 23-Sep 12:46 bacula-dir JobId 0: Fatal error: Error sending Hello to
> File daemon at "(hostname):9102". ERR=Interrupted system call
>
> Does every connection count as a "job"?
No, but it does count as a *connection*, so if you have concurrency on
the client set to 1, then any other access while a job is running will
be denied or time out. By the sound of it, you need to increase the
concurrency on the client.
> And do I have any recourse
> other than either wondering if one of the jobs will actually finish
> getting through the remaining 70+GB before the end of the day when the
> user will no doubt unplug his laptop and go home with it (leaving us
> still needing to get a full backup to run again most of the day
> tomorrow), or halting the bacula director entirely to "crash" the job,
> also probably leaving us unable to finish the full backup before
> quittin' time today and therefore still needing to do a full backup
> tomorrow again?
By the sound of it, you're pretty much between a rock and a hard place
there. If the Director can't connect to send a cancel, then you really
have no mechanism for killing just one of the running jobs.
--
Phil Stracchino, CDK#2 DoD#299792458 ICBM: 43.5607, -71.355
ala...@caerllewys.net ala...@metrocast.net p...@co.ordinate.org
Renaissance Man, Unix ronin, Perl hacker, Free Stater
It's not the years, it's the mileage.
------------------------------------------------------------------------------
Nokia and AT&T present the 2010 Calling All Innovators-North America contest
Create new apps & games for the Nokia N8 for consumers in U.S. and Canada
$10 million total in prizes - $4M cash, 500 devices, nearly $6M in marketing
Develop with Nokia Qt SDK, Web Runtime, or Java and Publish to Ovi Store
http://p.sf.net/sfu/nokia-dev2dev
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
