I have my home directory on my home machine, and NFS mount it at work:
dsl-tor:/ on /hosts/dsl-tor type nfs (rw,addr=126.96.36.199)
When I attempt to copy a large file to work, or access a large file
(for example, rpm -q -l -p apache-1.3.9-4.src.rpm) a series of the
following messages appear in /var/log/messages at work:
Feb 9 09:24:48 gnu kernel: nfs: task 4235 can't get a request slot
I must reboot the work machine (client) to clear the failure.
Others have reported the failure, but I've searched for days now for
a fix and come up empty handed. Trond wrote:
The messages themselves just mean that the RPC layer is congested, and
that a backlog of messages waiting to be sent is building up. Whenever
one of these messages times out, it issues the 'task xxx can't get a
request slot' error.
Please try to update the 'mount' utility, and then use something like
'mount -onolock server:/directory /mountpoint' since the userland NFS
does not support locking. If you still see the 'task xxx can't get a
request slot' message, please contact me directly via email as I am
the one responsible for the NFS changes in ac5.
I've been unable to find a new 'mount' utililty.
I'd like to help isolate the failure, but I don't know how to proceed...
Created attachment 106 [details]
Notes from my search for a fix
Tried turning off autofs, and using amd instead.
Continued to see this problem under stcok 6.2,
and the latest kernel: 2.2-16-3.
Every Saturday morning, when I back up my home directory
(on my home machine) to my work machine this failure occurs.
The system locks up, and requires a <RESET> and manual
fsck to recover.
assigned to johnsonm
Closing old 2.2 related bug. All our errata are now 2.4 based so this bug
doesn't contain information we believe is worth keeping