Usually I try to file a bug upstream and put an exception in the factory.

Sometimes I do a special recipe for just that package but I try to keep it all 
in the factory whenever I can.

Worst case I skip the packages... but that is very worst case

I haven't seen file conflicts like that in Cent or RHEL before so I will have 
to look into a fix.

--

Brett C. Smith
[email protected]
Sr Software Developer
Platform Deployment Technologies
(919)531-6635  -- x16635

________________________________________
From: Foresight-devel <[email protected]> on 
behalf of Michael K. Johnson <[email protected]>
Sent: Thursday, March 13, 2014 1:15 PM
To: Foresight Linux Development
Subject: [Foresight-devel] Tuned build cluster running faster

On Wed, Mar 12, 2014 at 06:50:42AM -0400, Michael K. Johnson wrote:
> Binary imports are now making progress.  However, we're now limited
> by available build hardware; we're importing only about 40 packages
> per hour in the current build farm, which adds up to about 15 days
> just to do the initial import.  That's in part because we haven't
> yet gotten the new build machines to deploy into the cluster,
> but in part because the build cluster was tuned for a few builds
> of source packages in which parallel make is very effective, not
> for many imports of binary packages in parallel.  So we'll look
> at both approaches to speed things up.

I finally had a chance to retune the build cluster a bit.  I guessed
at numbers and my guesses might not be optimal.  But it's sure trying
to build a lot more at once, and the machines are busier than they
were.  So far, I measured that in 2:42:45 (9765 seconds) we started
267 build jobs.  That measures out to a current average of 1.64
packages per minute.

When this restart finished calculating current state, it came up
with "updatebot.bot found 12566 packages to build" which would be
a bit less than 5.5 days to complete.


Now, we're finding some oddities from policy errors.

error: Error building recipe
bacula2:source=/f20s1.flnx.org@f:20sbeta/2.4.4_16.fc20-1[is: x86]:
Package Policy errors found:
file /var/spool/bacula2 added with conflicting contents: root:root
040770  != bacula:bacula 040750

error: Error building recipe
condor-cloud:source=/f20s1.flnx.org@f:20sbeta/0.1_7.fc20-1[is: x86]:
Package Policy errors found:
file /var/lib/condor-cloud/local_cache added with conflicting contents:
condor:condor 040755  != nobody:nobody 040755

error: Error building recipe
clamav:source=/f20s1.flnx.org@f:20sbeta/0.98_2.fc20-1[is: x86]: Package
Policy errors found:
file /var/lib/clamav/daily.cvd added with conflicting contents:
clamupdate:clamupdate 0100664
1c636ab26d52db1628816da0635c1f3a23d5eebc968b1faaa904d097d40760bb !=
clamupdate:clamupdate 0100644
1c636ab26d52db1628816da0635c1f3a23d5eebc968b1faaa904d097d40760bb
file /var/lib/clamav/main.cvd added with conflicting contents:
clamupdate:clamupdate 0100664
b02e221fc03a3ab651db740a503aa1eff3eec80d45afbb20d75df247bf3d3683 !=
clamupdate:clamupdate 0100644
b02e221fc03a3ab651db740a503aa1eff3eec80d45afbb20d75df247bf3d3683


Those seem like upstream bugs.

Brett, what have you tended to to with conflicts like this upstream
in the past?

Thanks!

_______________________________________________
Foresight-devel mailing list
[email protected]
https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
_______________________________________________
Foresight-devel mailing list
[email protected]
https://lists.foresightlinux.org/mailman/listinfo/foresight-devel

Reply via email to