From: Kumar Amit \(ext_TCS\) (Amit.Kumar.ext_TCS@ts.siemens.de)
Date: Wed Sep 03 2003 - 11:12:00 EDT
Hi Managers,
We are facing the Performance Issue after adding 2 Nodes to the production
Cluster.
Now it is a 4 node cluster. The same CVS checkout takes 1.5 times time now
and also
access of file system hangs in between.
The details of Compaq Cluster setup here are as below:
Machines : 4 COMPAQ DEC6600 DS20E
O.S : TruUnix5.1A and TruCluster 5.1A. with Patch Kit 4
CPU : All machines have 1 CPU and 1GB RAM.
CPU Speed & : Two Machines have EV6 (21264) 500Mhz Cache 4.0 MB
Type Two machines have EV6.7 (21264A) 667Mhz Cache 8.0 MB
Memory Channel : 2 CCMAB-AA connected thru Memory Channel Hub CCMHB-AA
MEMORY
Disk Storage : RA7000 storage array with dual HSZ70 controllers and Dual
SCSI hubs.
The controllers are configured in Transparent failover mode.
If 2 or 3 users are reading and writing more data at once, (ex. like tar and
untar of Big file of size more that 100MB), then it takes longer time that
before and user process hangs for moment in between.
After some kernel parameters changes, it is better, but still it takes
longer time.
We are suspecting that CFS is causing this problem and some kernel
parameters of this needs to be tuned, but we are not sure about it.
If you can point me where to look for improving the degraded performance
after adding 2 nodes, it will be a grat help to us.
Any help would be a great help.
Regards,
Amit Kumar
Unix System Administrator
Siemens TS RA D OC
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:34 EDT