On 23 Jan 2006, at 5:06 pm, Erez Zadok wrote:
Tim,
Does it happen on every machine with the same config, or just that one
machine? I want to try and eliminate a h/w flakiness as an issue.
If you get SEGV's, did anyone try to gdb a core and see exactly
where and
why the SEGV occurred?
I don't see from your amd.conf how the user is using direct maps,
and I only
see one map, not two.
No I couldn't work that either and have asked him about it. I'll get
back to you.
Direct maps aren't well working IIRC, due to
limitations of the linux kernel.
Can you run amd with full debugging on, so we can check if there's
anything
strange going on around the time of problems?
Yup, I'll suggest that.
Is the system in question heavily used at times? If the host has a
lot of
activity on it, the scheduler may not schedule amd often enough,
possibly
resulting in the "not responding -- still waiting" messages you
see. The
better answer here is to use amd in Autofs mode.
He claims that it's not under heavy load, but I'll pass on your
suggestions. Note that if you CC messages to [EMAIL PROTECTED]
(as I'm doing for this message), they will be automatically appended
to our bug track, and both the reporting user and I will be
notified. If you're interested in receiving notifications when a
Debian user files a bug against am-utils, it is possible for you to
be subscribed for receiving those reports - tell me if you're
interested.
As a start, hope this helps.
Thanks, it's a good place to start.
Tim
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
