Hi all,
I am experiencing the following problem:
I have two linux boxes connected first via a switched
100M-Ethernet using on-board NICs, and second by a private
1G ethernet cable connecting a DLink DGE550-T in each of the two
(in 32bit 33MHz PCI sockets). I am using dl2k v 1.17a on
linux-2.4.21 system (SuSE patched kernel linux-2.4.21-273 on the
NFS client, linux-2.4.21-286 on the NFS server).
However, NFS mounts over the 1G connection are often (but not alsways)
timed
out. Increased values for timeo and retrans as mount options do not
help.
It seems as if the mounts work if there is plenty of time (approx. 15 or
so
minutes) between them (I need to do 4 mounts all in all).
After each mount attempt, regardless if successful or not (i.e. timed
out)
I find a line like the following one in the syslog file of
the NFS server:
"May 16 11:57:36 chlorine rpc.mountd: authenticated mount request from
10.0.0.2:988 for /usr/local/httpd (/usr/local/httpd)"
Everything works fine when using the other NIC connected to the 100M net
on both machines.
Interestingly, when doing "route" at the NFS client, it takes a while
(say 7..8 secs) before printing the route for the 1G network connection
whereas the other routes come up much faster.
Any tips what that might be and how to overcome it?
Uwe
|
Similar Threads
Linear Mode
