NFS for clusters


Here are notes I've found useful for configuring reliable shared disk on a linux cluster...


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
Optimizing NFS Performance
tuning nfs I/O

While tuning nfs mount options, you can check on the transfer speeds doing simultaneous writes via:

for x in {1..10}; do echo "dd if=/dev/zero${x} bs=64k count=1000" ; done | xargs -n5 -P10 time

Run the command from the client, where nfs is mounted.

tuning nfs server threads

Of specific interest was the nfs server threads which can be monitored with:

watch -d "grep '^th' /proc/net/rpc/nfsd"

The first number is the number of threads available for servicing requests, and the the second number is the number of times that all threads have been needed. The remaining 10 numbers are a histogram showing how many seconds a certain fraction of the threads have been busy, starting with less than 10% of the threads and ending with more than 90% of the threads. If the last few numbers have accumulated a significant amount of time, then your server probably needs more threads.