Tr : NFSv4 idmapd memory consumption problem

[pierre_stephane.baton at alcatelaleniaspace.com]

Hello,

Just 1 line in /var/log/message when I mount a new nfs:

Sep  1 10:18:30 serverID kernel: SELinux: initialized (dev 0:a4, type 
nfs), uses genfs_contexts


nothing else.

Nothing when I umount.

(idmapd is demonised, with rpc.idmapd -f -vvv I can't see anything as it 
creates a new instance)



________________________________
Pierre-Stéphane BATON
DI - OI/IT
Consultant Trasys-OSI
Support informatique Unix/Linux
E01-H14-L44 - Tel :+32 (0)71 442797

Alcatel Alenia Space ETCA (DI/SYS)
Rue Chapelle BEAUSSART, 101
B-6032  Mont-sur-Marchienne



"J. Bruce Fields" <bfields at fieldses.org> 
31/08/2006 20:32

A
pierre_stephane.baton at alcatelaleniaspace.com
cc
nfsv4 at linux-nfs.org
Objet
Re: Tr : NFSv4 idmapd memory consumption problem






On Thu, Aug 31, 2006 at 04:23:20PM +0200, 
pierre_stephane.baton at alcatelaleniaspace.com wrote:
> More than you think, with Clearcase's vobs (with their #{|][# of virtual 

> file system) and others, if the changes of the mounted directories are 
> stored in the memory, the giga used is not realy strange. As those WS 
are 
> up for 17 days now, some informations has to  be logged by idmapd or be 
> deleted, or it should use all RAM memory.

OK.  Actually looking at the names of clients in that directory, I guess
you've seen some a few hundred thousand of them; divide that into a gig,
and it looks like we're leaking a few K on each mount.

That's weird.  Well, it'd probably be easy enough to reproduce:

                 while true:
                                 mount some nfs filesystem
                                 unmount it

and watch idmapd.

--b.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://linux-nfs.org/pipermail/nfsv4/attachments/20060901/66178bdb/attachment.htm