On Mon, Jan 16, 2012 at 12:56 AM, "Martin v. Löwis" <[email protected]>wrote:
> > What's being done to diagnose this further? > > The current diagnosis is that the disk driver reports SCSI errors on the > RAID controller. They don't get logged to disk, as the disk has > failed... > > Things go downhill from their, apparently causing a kernel hang > ultimately. > > The RAID controller itself reports that it is in "optimal" condition > (i.e. no hardware failures). > > I'm hesitant to perform a firmware update to the controller without > physical access in case the update fails. > > > 2 outages in the space of 3 or 4 days doesn't look to good. > I have full stats for http://www.python.org/ with latency since July 2011 with 1 minute discretion. I can export them into .cvs or somebody can help me with JS part to draw them at http://pydotorg.appspot.com > How does this correlate with pypi being moved to dinsdale? > > Perhaps not at all. The controller had a series of failures last summer > also, but then "fixed" itself somehow. It could be overheating of some > component (not necessarily the RAID controller), which could explain > why it occurs randomly, and in increased frequency after putting more > load on the machine. It is because of this guy - http://www.youtube.com/watch?v=tDacjrSCeq4 -- anatoly t.
_______________________________________________ pydotorg-www mailing list [email protected] http://mail.python.org/mailman/listinfo/pydotorg-www
