Description of problem: Binary build configs randomly fail with this error: The pod for this build already exists and is older than the build Error: # oc get builds builds/pega-jboss-22 Docker Binary Complete 17 hours ago 1m21s builds/pega-jboss-23 Docker Binary Error (BuildPodExists) 15 hours ago builds/pega-jboss-24 Docker Binary Complete 15 hours ago 1m21s The issue for this error exists in the Upstream and appears to be resolved with the workaround is: To delete the BC before creating the new one. Deleting a BC will delete all associated builds, thus resetting the build count to 0. The upstream github link for this issue: https://github.com/openshift/origin/issues/9301 Actual results: Build [oc start-build] needs to be triggered multiple times in order to have a successful build. Expected results: Every build started with [oc start-build] for binary should be completed with the proper build number.
I don't think the issue you referenced is related to your problem. Did you previously have a buildconfig named pega-jboss? We have recently seen this problem when the clocks associated w/ all the masters were not synced. Can you ensure your clocks are synced?
Hi Ben I asked this question in the users list a couple of weeks ago. Unfortunaltly I wasn't able to find much in the logs and the issue is not as reproducible as I thought. But it is happening very frequently now. Clock sync issues sounds possible. We have had problems with ntpd in our RHEL nodes before.
Well you might be onto something. ntpd was down in one of the masters and the clock was off by one second. I will have to keep an eye on the builds and ntpd to confirm. This will lead to another ticket to find out how ntpd is stopping.
For completion I am not modifying or recreating the failing build configs in any way. The linked origin issue is not relevant.
*** This bug has been marked as a duplicate of bug 1547551 ***