You probably hit a known bug, See
6559232 sometimes failed to remove file by NFS UDP during testing
multiple NFS connections <http://monaco.sfbay.sun.com/detail.jsf?cr=6559232>
- Zeeshan
On 09/18/09 07:21, Tom Chen wrote:
Hello,
We are doing NICDrv test to test our GLDv3 Network driver and test03 (nfs
corruption) always fail in the nfs udp corruption tests on some newer, faster,
servers but it always pass on some slower machines. We are quite puzzled.
stdout| Setup Summary
stdout| ======================================
stdout| PRODUCT : nicdrv
stdout| CLIENT_INT : 192.168.43.79
stdout| SERVER_INT : 192.168.43.80
stdout| SESSIONS : 10
stdout| TIME : 3600 sec
stdout| TRAFFIC_TYPE : bi
stdout| PROTO : udp
stdout| RUN_ON_ERROR : no
stdout|
stdout|
stdout| Corrupt Status
stdout| ======================================
stdout| FAILED: 18 of 20 sessions finished
We checked internal chip register status and did not see any packet dropped. However, the
"nfsstat" command shows that a lot of retransmissions and timeouts. For
example, there are 66639 re-trans and timeouts in a total of 1854131 calls. We do not
know where the packets got dropped? Why timeout happened? Is this the reason why test03
failed?
Tom
#nfsstat
......
Client rpc:
Connectionless:
calls badcalls retrans badxids timeouts newcreds badverfs
1854131 0 66639 6803 66639 0 0
timers nomem cantsend
69137 0 0
bash-3.2# uname -a
SunOS galilei 5.11 snv_121 sun4v sparc SUNW,T5240
_______________________________________________
networking-discuss mailing list
[email protected]
