NFS / ZFS Issues

M

magickan

Hello All

Hopefully someone can help. I seem to be running into an issue fairly regularly with transfers from my linux workstation to my freebsd file server. After a while the server seems to become completely unresponsive via network. No icmp, no http, no ssh etc. The server is still up and if i connect to it locally, its fine, but i cant get the network back up except by a reboot. This seems to be when transfering ~ 20Gb or more but im not sure if this cumulative. The zfs is shared via nfs using the inbuilt zfs nfs commands.

The major part of the problem for this is I'm not sure where i should be looking in order to troubleshoot the issue. Which logs etc. I am considering leaving a tcpdump running on it for analysis and then transferring to see if that helps, but i was wondering what inbuilt system tools i can use to work out whats going on.

Any help with this is will be appreciated.
 
Next time make a tcpdump on the NIC you are using for NFS

Code: 
tcpdump -i em1 -n -c 100

If you see INPUT/OUTPUT Errors:

Code: 
6:52:13.244787 IP 10.0.0.3.1249424650 > 10.0.0.23.2049: 4356 write [|nfs]
06:52:13.244794 IP 10.0.0.2.1825385927 > 10.0.0.23.2049: 176 write [|nfs]
06:52:13.244822 IP 10.0.0.23.2049 > 10.0.0.3.1249424650: reply ok 36 write ERROR: Input/output error
06:52:13.244911 IP 10.0.0.7.2129762546 > 10.0.0.23.2049: 144 write [|nfs]
06:52:13.244919 IP 10.0.0.23.2049 > 10.0.0.2.1825385927: reply ok 36 write ERROR: Input/output error
06:52:13.244932 IP 10.0.0.23.2049 > 10.0.0.7.2129762546: reply ok 36 write ERROR: Input/output error
06:52:13.245036 IP 10.0.0.3.1249424651 > 10.0.0.23.2049: 4356 write [|nfs]
06:52:13.245042 IP 10.0.0.2.1825385928 > 10.0.0.23.2049: 176 write [|nfs]
06:52:13.245067 IP 10.0.0.23.2049 > 10.0.0.3.1249424651: reply ok 36 write ERROR: Input/output error
06:52:13.245072 IP 10.0.0.23.2049 > 10.0.0.2.1825385928: reply ok 36 write ERROR: Input/output error
06:52:13.245161 IP 10.0.0.7.2129762547 > 10.0.0.23.2049: 144 write [|nfs]
06:52:13.245167 IP 10.0.0.2.1825385929 > 10.0.0.23.2049: 176 write [|nfs]

If your errors looking like that then you hit a bug.

Upgrade to RELENG_8 or include the patch by hand.

The problem and the solution is discribed -> http://lists.freebsd.org/pipermail/freebsd-questions/2010-March/213941.html
 
You must log in or register to reply here.
Back
Top