I have a server that hosts images and allows access to one directory for one client over NFS.
The directory contains about 70K images and is 12GB big.
When I execute commands like cp
or rm
on the server that hosts the NFS directory performance is stellar, but when execute the cp
or rm
from the client on the NFS directory commands sometimes get really slow where the real time
is slow and user
and sys
times are fast:
real 1m33.409s
user 0m0.000s
sys 0m0.000s
My assumptions are:
- The directory gets somehow locked thus causing the long wait
- The connection between the two servers is somehow not reliable
Here's a dump from the client's nfsstat -c
Client rpc stats:
calls retrans authrefrsh
3018431 1 3018571
Best Answer
You should try the
noatime
option when mounting the remote filesystem in the client.From
man 8 mount
In my case, it improved write performance a lot.