|
|
Boost Testing : |
From: Rene Rivera (grafikrobot_at_[hidden])
Date: 2007-11-20 10:33:55
Markus Schöpflin wrote:
> John Pavel wrote:
>> My expectation of the results changing every 2-3 hours seems
>> misconceived. Can anyone enlighten me?
>
> I'm asking myself the same question. The report info page currently tells me:
>
> previous run: Mon Nov 19 14:17:30 UTC 2007
> current run: Mon Nov 19 22:18:07 UTC 2007
>
> The time of the current run matches the time of the report found at the top
> of the report info page.
>
> At the time of writing this mail, date -u report Tue Nov 20 12:05:59 UTC 2007.
>
> So, each run takes about 8 hours. That would mean that the next run should
> have happened at 06:18 today. But now it's already 6 hours later, and still
> the results from yesterday are showing.
>
> So what do these timestamps actually mean? The start of the report
> generation cycle? Or the end of it? And is each cycle really taking 8
> hours, or is the generation of the release branch results included here?
This is just the trunk generation... And the times indicate the start of
the run. So it would mean that the results might show up at
current+2*delta. Or another way, if you look at the results shortly
after they are uploaded the date will be current+delta.
> Also, the report generation machine shows a load average of 0.15/0.03/0.01,
> which doesn't seem right to me if the reports are really regenerated
> continuously.
Hm, there's a rest period of five minutes between each run (to give me a
chance to stop it and to OS updates as needed). But it's likely that the
load is not high because two large chunk of time are taken up by steps
that don't seem to take much CPU. Which would be generating the zip
archive, and uploading it.
-- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo