rpc.lockd resource starvation

[Eric van Gyzen]

Dan Nelson wrote:
> In the last episode (Jan 15), Eric van Gyzen said:
> > I'm running 5.1-RELEASE on my NFS server and my ~50 NFS clients.
> > Over a period of a few weeks, the rpc.lockd daemon on the NFS server
> > will consume all the privileged udp ports and start using
> > high-numbered ports.  With no available privileged udp ports, the
> > server is unable to mount NFS shares from other machines.  (There are
> > probably several other unfortunate consequences of which I am not yet
> > aware...) Is this behavior expected from rpc.lockd, or might it be a
> > bug (or just me breaking my systems again)?
>
> wow
>
> I think you just told me why my two busiest NFS servers had to be
> rebooted a few months ago (one with 440 days of uptime :( ).  Does the
> mount fail with "mount: Can't assign requested address"?

Yep.

> If so, it
> also happens on 4.x servers.  Currently, they have 214 and 109 open
> reserved ports (after 102 and 73 days uptime, respectively), and I'm
> betting there are no more than 5 files actually locked on either
> system.  I wonder if it's just not closing sockets when it's done with
> them?

Sounds like a reasonable guess to me.  I doubt there are more than 424 
locked files on my server (the default number of available low-numbered 
ports).  I could be wrong, though...I wonder how I could determine which 
files it currently holds locked...

Eric

-- 
Eric van Gyzen                        Sr. Systems Programmer
http://www.stat.duke.edu/~vangyzen/   ISDS, Duke University