Community
Participate
Working Groups
The Eclipse Platform recently moved what were previously cron jobs to Hudson. We schedule these as before, such as "10 oclock on Tuesday" and typically just use the time "now" as the timestamp for the build. Such as we snag the "now" time with RAWDATE=$( date +%s ) when the build first starts. BUT, even if we do that as the first line of a "shell script" build step, it is already 10:01 instead of 10:00. I am curious if it is known why this is? Does it really take a Hudson job a minute to "startup"? Or, does Hudson only check for periodic jobs at the "end of a minute"? Or is is some amount of random time depending on what else Hudson is doing? In a sense, kind of minor, since the "timestamp" is sort of arbitrary identifier. but, looks confusing to those used to seeing the "time of the build" as 10:00. So, I am wondering if it is known why it is a "minute late", then we might be able compenstate in a predictable way. But if it is sort of a "random amount of time", then we'd would not be able to. Thanks,
I meant to also give our Platform bug. It is bug 496345. I have added your "BUILD_ID" to a debug statement to see if it is anymore accurate "to the minute".
(In reply to David Williams from comment #1) > I meant to also give our Platform bug. It is bug 496345. > > I have added your "BUILD_ID" to a debug statement to see if it is anymore > accurate "to the minute". It was not more accurate. Still a minute "off". See bug 496345 comment 5