LinuxQuestions.org - [SOLVED] NFS clients regular high getattr

Hello,

Since last Friday I experience a regular high rate of getattr from two of my NFS clients. This look like this:

Attachment 11043

On both servers Cacti reports the same stats.
I have no idea where it come from, but while searching for an answer, I found that the syslogd process had a high i/o wait, around 80%. So I did a strace of the process and noticed the following, non stop lightning speed occurring.

Attachment 11044

When I kill the syslogd, the i/o wait disappear but not my getattr rate.

The NFS server stats are the following :

Code:

Server nfs v3:

null        getattr      setattr      lookup      access      readlink

152      0% 1783936589 55% 35924507  1% 259828607  8% 526889444 16% 8332      0%

One the clients:

#1

Code:

Client nfs v3:

null        getattr      setattr      lookup      access      readlink

0        0% 25779368 52% 512655    1% 5714549  11% 6687009  13% 234      0%

Code:

Client nfs v3:

null        getattr      setattr      lookup      access      readlink

0        0% 27243070 53% 107335    0% 7549252  14% 7989644  15% 476      0%

Those two clients are Apache webservers under RedHat Enterprise Linux 5.8 with NFS mount on another RHEL machine.

Thank you.