|
|
Boost Testing : |
From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2006-02-17 02:32:22
Jim Douglas writes:
> Not that it is a major issue, but I find it increasingly frustrating
> to have to wait at least 4 hours for my regression test results to
> show up on the website. Is there a sound justifiable reason for
> this?
It's not really that sound or justifiable, but building the reports
involves processing huge volumes of data, and two and a half hours is
just what it currently takes to regenerate the pages. It appears to
you as something close to a 4-hour delay because most of the time when
your (or somebody else's) results are uploaded to the FTP, the reports
regeneration is already in process and the new results have to wait to
be picked up by the next cycle.
At some point we've optimized things here and there and brought the
cycle under 15 minutes, but obviously the increased volume of data
pushed the times back to hours. I totally agree that it is a major
inconvenience, and resolving it is on our short-term TODO list.
> There seems to be an inconsistency with the clock setting. If I
> access the site and get the "unpacking" message, my assumption is
> that the report is being updated (is this correct?).
Yes.
> If I check the time on the report immediately afterwards it appears
> to be 2 hours behind UTC.
Because it's the time of the _beginning_ of the new cycle.
-- Aleksey Gurtovoy MetaCommunications Engineering