ps output staggered / choppy

[Mark McHugh]

I have 2 servers configured exactly the same, that sit on the same subnet, same HW, etc etc. On the first server, if I enter the command ps -eaf I get a smooth scrolling output. On the second server with the same command I get a choppy output, with random pauses after a few pages of process output, not at all smooth. While they both have a similar number of processes and workload etc, the command takes 10 or 12 seconds to come back whereas on the first box it is always less than 5. Any thoughts?

Much appreciated.

[Zafar A. Mohammed_1]

Is the network delay, or more processes.

Thanks
Zafar

[Paula J Frazer-Campbell]

Mark

No two server are exactly the same as data is not in the same place on the disks, the same memory chunks are not being used to carry out the same task eyc-etc.

- have you got glance installed if so fire it up and watch the ps -eaf also TUSC can be used to see where the delay is occuring.

A 60 trial of glance in on the HP-UX CDs.

Paula

[John Dvorchak]

It could be any number of things related to and not related to ps -eaf. That may just be the sympton of something else wrong. If top, glance, uptime etc all show similar values of total utilization then you could be looking at a LAN related problem. You will have to tell us a little more about the state of the boxes before I would even venture to guess.

Are there any other symptoms? I am assuming that you telnet-ing to them and not on the serial consoles. If all seems the same, look at the LAN settings and just confirm that they are both set the same. If 10mb there is not a lot to do. If they are on 100mb check that both have the same speed and duplex settings on the NIC's. You can us lanscan to get the ID's and then lanadmin to check the settings.

[Mark Greene_1]

Check the kernel settings via sam to see if there are differences. Try a "ps -ef |wc -l" on each system to see if there is a large disparity in the number of processes running. Do "ipcs -ca|wc -l" to see if there is a big difference in semaphore useage. This would tend to indicate a large difference in the amount and way in which memory is utilized. In fact, run "vmstat 2 20" to see if there is an excessive amount of paging going on.

In sam you can also check the top cpu processes. The systems may be utilizied just differently enough to explain what you are seeing.

One more thought, see what the NICs are set to. If one is 100 full and the other is 100 half; or worse, if the problem one is set to 100, full, auto-negotiate, but the switch only sees it as 100 half, then the retranmits are going to definitely affect communications performance.

HTH
mark

[Oleg Zieaev_1]

Mark,
if there is a delay, the systems are not 'exactly the same'.
You might concider checking the settings of the lan cards on the second server to verify you are not using AUTO mode.
Also using glance or netstat -in check the load for the interfaces. Most likely there is a problem on the network level (you might require diagnostic run on the switch port you are connecting to).

Hope this helps
-0leg