Hi,
I have a problem with the long term performance of NFS.
After some days (or weeks) of uptime, the following messages set in on
the NFS CLIENT ("ficus5"):
Jun 4 13:09:25 ficus5 kernel: nfs: server linde not responding, still trying
Jun 4 13:09:35 ficus5 kernel: nfs: server linde OK
Jun 4 13:19:36 ficus5 kernel: nfs: server linde not responding, still trying
Jun 4 13:19:36 ficus5 kernel: nfs: server linde OK
Jun 4 13:28:58 ficus5 kernel: nfs: server ficus7 not responding, still trying
Jun 4 13:29:24 ficus5 kernel: nfs: task 35715 can't get a request slot
Jun 4 13:29:33 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:29:38 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:29:49 ficus5 kernel: nfs: server ficus7 not responding, still trying
Jun 4 13:29:49 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:29:50 ficus5 kernel: nfs: task 35728 can't get a request slot
Jun 4 13:29:56 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:30:06 ficus5 kernel: nfs: server ficus7 not responding, still trying
Jun 4 13:30:08 ficus5 kernel: nfs: task 35750 can't get a request slot
Jun 4 13:30:08 ficus5 kernel: nfs: task 35749 can't get a request slot
Jun 4 13:30:09 ficus5 kernel: nfs: task 35754 can't get a request slot
Jun 4 13:30:36 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:30:47 ficus5 kernel: nfs: server ficus7 not responding, still trying
Jun 4 13:30:47 ficus5 kernel: nfs: server ficus7 OK
Jun 4 13:30:52 ficus5 last message repeated 3 times
...
At the same time, all operations with nfs file systems slow down and make
the client unusable. Unmounting or restarting nfs doesn't help.
Note that server "ficus7" is another linux system, but "linde" is a HP
(running HP-UX 10.20). At the same time, other clients can access these
servers as always. The sick client can only be cured by a REBOOT (AFAIK).
The symptoms may reappear after days or weeks of normal uptime.
Hardware: Dual PPro200 on S1662D 440FX PCI-ISA board
Adaptec AHA-2940A Ultra SCSI host adapter
SMC-Ultra ethernet card
The problem was already present when running
RedHat 4.2 / kernel 2.0.3x-SMP
and unfortunately persists under
Debian 2.1 / kernel 2.2.{4,9}-SMP
Since all our linux boxes are running SMP kernels, I don't know whether
the problem is related to SMP.
How to get further diagnostics?
Need fancy mount options?
Any ideas?
Burkhard
----------------------------------------------------------------------
[EMAIL PROTECTED] Physics Institute, Humboldt University
fax: ++49-30 2093 7628 Invalidenstr. 110
phone: ++49-30 2093 7646 10115 Berlin, Germany
----------------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
