Re: [Evolution] Stalled evolution-mail process



On 11 Aug 2001 09:41:07 +1000, David Luyer wrote:
On Fri, Aug 10, 2001 at 04:35:28PM +0200, Frederic Crozat wrote:
On 10 Aug 2001 23:03:21 +1000, David Luyer wrote:
A 'D' state could quite easily be the kernel's fault.  Does 'dmesg' have
anything interesting in it?

Nothing :(((

How about in ps -axo stat,pid,wchan ... what's the wchan in the stuck
process?
Just got 2 evolution-mail threads stalled :

ps -axo stat,pid,wchan | grep 2373
D     2373 down_write_failed
ps -axo stat,pid,wchan | grep 2376
D     2376 down_write_failed


Also you might want to check your mount options for the NFS, maybe
something like soft,intr,timeo=14 could be added to make it timeout
and abort NFS requests better if anything is going wrong with the
client-server communication (which seems to happen even over direct
switched ethernet at times with many Linux kernel revs).

Again, .nfs* files are nothing to do with Evolution either.

Well, I'm not so sure, since after each "Receive new messages", a new
.nfs file appears (probably generated by deleting movemail files) and
appears as used (deleted) by evolution-mail (seen in lsof).. And I'm the
only one accessing this file.. I'm wondering if there is several threads
trying to access the same file..

OK, when I wrote they were nothing to do with Evolution what I should
have written is "caused by Evolution but they're not Evolution's
responsibility to track".  It's up to nfs/the kernels (of the client and
server) to track these things - Evolution should have no idea that every
time it deletes a file via NFS that is held open at the time it results
in a .nfs* file.

Ok
-- 
Frédéric Crozat
MandrakeSoft





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]