On Mon, Jan 18, 2010 at 8:10 AM, Adrian Schröter
Am Montag, 18. Januar 2010 14:51:03 schrieb Jon Nelson:
I continue to have build loops (not actual loops) I can't explain. They do eventually run themselves out, but sometimes I see packages rebuild 30, 40, 100 or more times. In some cases, there are explanations for why they might rebuild a few times, but not 10 times much less over a hundred.
For example, one package in x86_64 has rebuilt (around) 47 times in the last 3 days. The jobhistory shows only 'rebuild counter'.
We deliver the obs in default config that i586 and x86_64 packages should always have the same release number. This means, if it builds for one architecture, the other one gets triggered as well.
If you done like this, define this in your BSConfig.pm to have seperate pools:
our $relsync_pool = { ... "i586" => "i586", "x86_64" => "x86_64", ... }
I made that change but today I was getting complaints of packages rebuilding over and over and over. The build history showed the same srcmd5, same revision, and builds every 2-5 minutes, until I shut down and restarted the scheduler. I'd love to be able to correlate the timestamps but why don't the logs have timestamps? How to I tell the ruby/perl/whatever logging system to log timestamps (without resorting to logging to syslog)? Anyway, it looks like glibc, glibc-common, and glibc-headers: !glibc !glibc-common !glibc-headers meta change, start build were the packages triggering the rebuild. However, after rebuilding, it would rebuild again, and for the same reasons! Sigh. What might be going on here? -- Jon -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-buildservice+help@opensuse.org