NFS stale error on Clustering setup!!!!
I am setting up a HA cluster with following
One Linux File Server with NFS share of Qmail with
vpopmail,courier imap and Mysql , One postgresql
Two HA nodes , node1 and node2 running in Active
Passive setup with NFS shared directories from Linux
Totally three machines are there in the setup.
The server was setup and made active . There are some
900 client users connecting to this server for POP3
mail fetching ,
and some 30 concurrent users connecting to postgres
Some 300 users concurrently connect and mail at Peak
time with an average of 100 MB size of traffic .
The node1 configuration is a PIII 733 MHz with 512 MB
DDR Ram .
The node2 configuration is same as node1.
The Linux File server configuration is an HP Machine
with SCSI HDD. having PIII 733 MHz processor and 750
I faced the following problems when server was made
1) During Peak time of mail traffic, qmail - smtpd and
qmail - pop3d is not responding.,.No outlook clients
can establish a connection with mail server.
Only after restarting qmail, mail processing
When qmail is restarted, the functionality becomes ok.
This happens only during peak traffic.
2) NFS Stale error also comes during peak traffic,
when NFS requests becomes high.
On restarting the node1 of HA Cluster , NFS becomes
3) Memory utilisations shoots to 99% at peak times.
Please let me know what fine tuning has to be done on
NFS inorder to manage the load. Basic NFS fine tuning
has been already done.