Hey guys, so we've almost got epel builds back online. Right now there's a communication problem between our plague server and our ppc builders. I've put a ticket in to the network team. As soon as that's done epel builds should be working again. Right now if you submit a build. It will just queue until that PPC builder is ready again.
Also with our new keys, the plague client is failing. See:
https://www.redhat.com/archives/fedora-devel-list/2008-August/msg00996.html
for a description of the problem and its resolution. I believe a new plague client will be built soon to solve this issue.
-Mike
Mike McGrath wrote:
Hey guys, so we've almost got epel builds back online. Right now there's a communication problem between our plague server and our ppc builders. I've put a ticket in to the network team. As soon as that's done epel builds should be working again. Right now if you submit a build. It will just queue until that PPC builder is ready again.
I don't know if this is important (or related to the above) but it is some sort of bug: I queued an awstats build just now. A few *seconds* after I submitted it, the buildsys web interface showed:
x86_64: hammer3.fedora.redhat.com Status: done/done Build Time: 5 minutes
i386: hammer3.fedora.redhat.com Status: done/done Build Time: 24 minutes
noarch: xenbuilder2.fedora.redhat.com Status: running/downloaded
Clearly, the x86_64/i386 figures are bogus. I've got no idea where they're coming from (it doesn't appear to be data leakage from one of the other jobs, because the figures don't match) but since this is a noarch package they shouldn't be there at all.
On the positive side, the noarch build ran correctly and built, so thanks to everyone who has worked hard to get these systems up and running again.
Final (very minor) point: can we have the web redirect from / restored? http://buildsys.fedoraproject.org/ just gives an empty page right now.
Tim
On Sun, 24 Aug 2008 08:28:23 +0100, Tim Jackson wrote:
I don't know if this is important (or related to the above) but it is some sort of bug: I queued an awstats build just now. A few *seconds* after I submitted it, the buildsys web interface showed:
x86_64: hammer3.fedora.redhat.com Status: done/done Build Time: 5 minutes
i386: hammer3.fedora.redhat.com Status: done/done Build Time: 24 minutes
noarch: xenbuilder2.fedora.redhat.com Status: running/downloaded
Clearly, the x86_64/i386 figures are bogus. I've got no idea where they're coming from (it doesn't appear to be data leakage from one of the other jobs, because the figures don't match) but since this is a noarch package they shouldn't be there at all.
There are more symptoms:
Build jobs mention hammer1,2,3 and ppc1 but these are not listed by "plague-client list_builders".
One can see that the build master has restarted at build-job #1 while old data from jobs 39770 to 39844 are still available in the jobsdb. Database corruption?
"Successful builds" web page gives "Error: Invalid job status".
Michael Schwendt wrote:
One can see that the build master has restarted at build-job #1 while old data from jobs 39770 to 39844 are still available in the jobsdb. Database corruption?
"Successful builds" web page gives "Error: Invalid job status".
I noticed these too - they are filed in the ticketing system.
Tim
On Monday 25 August 2008 09:34:32 am Tim Jackson wrote:
Michael Schwendt wrote:
One can see that the build master has restarted at build-job #1 while old data from jobs 39770 to 39844 are still available in the jobsdb. Database corruption?
"Successful builds" web page gives "Error: Invalid job status".
I noticed these too - they are filed in the ticketing system.
I fixed it yesterday afternoon.
The cvs checkout for the web front end was from head not the 0.4 branch. it had a few extra status's in it.
Dennis
On Mon, 25 Aug 2008 10:24:25 -0500, Dennis Gilmore wrote:
On Monday 25 August 2008 09:34:32 am Tim Jackson wrote:
Michael Schwendt wrote:
One can see that the build master has restarted at build-job #1 while old data from jobs 39770 to 39844 are still available in the jobsdb. Database corruption?
"Successful builds" web page gives "Error: Invalid job status".
I noticed these too - they are filed in the ticketing system.
I fixed it yesterday afternoon.
The cvs checkout for the web front end was from head not the 0.4 branch. it had a few extra status's in it.
Precisely, it can be found in the 0.4.5 branch/fork. Head is 0.5.0 devel.
epel-devel@lists.fedoraproject.org
-
Dennis Gilmore -
Michael Schwendt -
Mike McGrath -
Tim Jackson