|
Ted Pavlic wrote:
>
> Do an lsof and see what pulse is doing with all of those handles.
That's the interesting thing, they don't show up anywhere in lsof.
Shutting down pulse/lvs/nanny also doesn't get rid of them. Rebooting
the machine or just letting them slowly time out is the only way to get
rid of them. I'm not sure how that could be since it seems like once
you've killed off the offending process, the kernel should de-allocate
the resources...
> Maybe it has something to do with the communication between the two
> directors. Perhaps the directors try aggressively to talk to each other and
> end up timing out and only the communication channels which actually are
> being used survive. (after the timeouts)
That's my guess, because you can actually watch the values in
/proc/sys/fs/file-nr jump if you cut off the primary and let the backup
take over, right about when pulse realizes that the primary box is no
longer responding, and vice-versa when the primary comes back to life.
> Anyway -- it'd be interesting to see what lsof says pulse is using those
> filehandles for.
More interesting to find out it doesn't see them...
"Curiouser and curiouser," Alice said.
Since all evidence points to this being specifically RedHat-related, I
wonder if I should take this thread over to a RedHat-HA specific list
someplace? Or is this still interesting to the LVS community in
general?
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
With Microsoft products, failure is not Derek Glidden
an option - it's a standard component. http://3dlinux.org/
Choose your life. Choose your http://www.tbcpc.org/
future. Choose Linux. http://www.illusionary.com/
|
