Description of problem: copying files from a nfs mount to the local filesystem results in a oops. As a result the machine completely hangs and has to be rebooted manually. Version-Release number of selected component (if applicable): kernel 2.6.12-1.1372_FC3 How reproducible: always Steps to Reproduce: 1. ssh from source system (machine a) to target system (machine b) 2. On machine b: stop iptables 2. On machine b: mount nfs share from machine a (also FC3 with same kernel ver) 3. On machine b: copy files from machine a to local file system 4. On machine b: Oops.... Actual results: Kernel oops Expected results: copy should work Additional info: both machines run the latest FC3 kernel and updates.
Created attachment 118195 [details] Oops output from machine b, captured from killed ssh session
what is strange about this is the fact the server is failing the messages with 'call_verify: RPC call version mismatch' which means the client is either sending crap or the message is getting corrupted... would it be possible to get an binary tethereal trace of this? something like 'tethereal -w /tmp/data.pcap host server' then post the b2zipped data file.
Meanwhile I have upgraded to kernel 1376 and today to 1378 and have not seen the problem since. The nfs (and others btw) problem seemed to be hardware related, I had to remove one of the memory DIMMs to prevent te system from randomly crashing after several hours. I now runs stable again. This case can be closed, not a nfs bug.