Messages in this thread | | | Date | Sun, 10 Jan 1999 22:33:35 +0000 (GMT) | From | Chris Evans <> | Subject | 2.2.0p6: NFS funnies |
| |
Hi,
I was stress testing the NFS client against userland NFS server across localhost.
I don't think NFS over TCP is supposed to work too well yet, but I tried it anyway. In case anyone is interested I did
mount -t nfs -o nolock,rsize=16384,wsize=16384,intr,tcp localhost:/usr/src /mnt
Then to stress test, "cd /usr/src/linux; make -j zImage"
Things didn't go too well, I got all variety of dangerous sounding messages. Here is a sample.
Jan 10 22:14:47 localhost kernel: RPC: empty TCP record. Jan 10 22:14:47 localhost kernel: rpciod: active tasks at shutdown?! Jan 10 22:14:47 localhost kernel: nfs_revalidate_inode: drivers/pci ^^^^^ lots of getattr() failed, error always seems to be -5
I also got a RPC: wot no daemon? and some other messages but at this stage I had killed the loggers (along with everything else!)
Once I got the system to stop thrashing and killed everything, all getattr() calls to the still mounted filesystem failed despite an idle system and plenty of memory.
I can happily repeat the exercise if anyone wants.
Note that TCP NFS seemed to work pretty well for less loaded conditions (and with rsize=wsize=8192).
Cheers Chris
P.S. After all my NFS expermiments about 6Mb memory repeatably goes "missing" - any ideas?
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |