[Systems] aslo-web went down

Bernie Innocenti bernie at codewiz.org
Tue Dec 8 15:20:33 EST 2009

El Sat, 05-12-2009 a las 06:22 -0600, dfarning at sugarlabs.org escribió:
> Yes the freeze is related to the interaction of duplicity,
> gpg, and apache.  The machine got very sluggish during the
> back up.

Ugh, it occurs to me that *all* the VMs are probably running
their daily cronjobs at the very same time.

We should definitely fix it by shifting the cronjobs on
each VM by 1h.

> At one point duplicity was using over 3G of cache.

That's not really an issue: by design, the kernel will
use all free memory as a buffercache. A running backup
triggers this to happen quickly.

Memory in use by running applications will generally not
be evicted for this purpose, although a few cold pages may
be swapped out occasionally.

> The apache processes started building up, making the
>memory situation worse. I killed apache, and the memory
>held at between 80 and 60M free.

Hmm... whatever killed treehouse must be more powerful
than a simple cpu and ram overload: the kvm guests are
not supposed to be able to kill their host under any
condition, and anyway treehouse doesn't seem to be coming
back to life even after a power cycle.

   // Bernie Innocenti - http://codewiz.org/
 \X/  Sugar Labs       - http://sugarlabs.org/

More information about the Systems mailing list