topic Re: VAx 7600 in Operating System - OpenVMS

VAx 7600

vmsserbo — Thu, 29 Jun 2006 11:30:52 GMT

Users are complaining that the system is slow. It is clustered. We have 3 of them. What command can I use to see what is going on and what do I look for. I know about the monitor system.

Maybe it is a network issue?

Thanks!

Re: VAx 7600

Robert Gezelter — Thu, 29 Jun 2006 11:35:08 GMT

Miles,

There are many reasons why the system could be perceived as slow (every time that I think I have seen them all, somebody shows me a new way to create a performance problem).

In the instant. the MONITOR command will show a variety of information about the running enviornment. Often MONITOR PROCESS/TOPCPU will show if you are suffering from a CPU saturation.

There are also many other posibilities, but that will do for a start.

- Bob Gezelter, http://www.rlgsc.com

Re: VAx 7600

Edwin R. van der Kaaij — Thu, 29 Jun 2006 14:41:07 GMT

Hello,

$ MONITOR DECNET - shows the packets
$ MONITOR CLUSTER - shows each system

$ MC NCP
NCP> show counters node

As mentioned before the MONITOR tool
is powerful enough to get a good feel
for the performance.

Ed.

Re: VAx 7600

vmsserbo — Thu, 29 Jun 2006 14:44:24 GMT

What am I looking for?

Re: VAx 7600

Jan van den Ende — Thu, 29 Jun 2006 14:50:37 GMT

Miles,

the quickest way to see if the system can cope with the workload is
If the number of COM processes is zero half of the time or better, AND there are NO processes in RWxxx states, THEN for sure the SYSTEM is doing fine.
Then, MONITOR DISK/ITEM=QUE running for several minutes, if AVERAGE is less than 1, there is little disk-IO problem.
Too heavy loaded memory will have MONITOR IO showing substantial PAGE WRITE IO.

If any of these occur, that is the factor to be addressed.

If all of this is NOT happening, then your issue is with network, or anything between the keyboard and the VMS system (or maybe bad applic design).

hth

Proost.

Have one on me.

jpe

Re: VAx 7600

Edwin R. van der Kaaij — Thu, 29 Jun 2006 14:52:44 GMT

in your clustered environment it is important to fine any kind of resource that is getting depleted.

in your DECNET traffic it's important to see if you are receiving and sending packets in a regular fashion.

the MONITOR tool shows you how and by whom resources are being used. HELP MONITOR shows you what you can look at.

are there specific times of day that things seem slow? are your users physically very far from the system? can their perception be funded by usage numbers?

performance management is an art that needs many hours of collecting data and implementing minor changes to parameters :o)

Re: VAx 7600

Arch_Muthiah — Thu, 29 Jun 2006 15:50:47 GMT

Miles,

$ show system -- will show all the processes and its status, on the basis of the process status, you can take the necessary actions.

also you can use MONITOR command to see various acitivities related with LOCK, DISK IO, process, states, cluster, etc...

you can use SYSMAN command also to see all the activities going on cluster wide.

Archunan

Re: VAx 7600

John Abbott_2 — Fri, 30 Jun 2006 01:17:21 GMT

Hi Miles,

7600... great machines :-)

I always find the $ MONITOR MODES is a useful command to start with, see if you've got any "idle time" on the system.

Next $ MONITOR PROCESS/TOPCPU to see if there are any DB/SERVER type processes that are running at or near to 100% this might indicate a bottle neck in the DB/server.

MONITOR SYSTEM as Jan mentions. All the posts here are good for detective work.

I'm sure it's all quite now :-) so have a "play" with the commands posted by all, I'm sure everyone will be willing to reply to any additional questions you might have.

Regards
John.

Re: VAx 7600

Zeni B. Schleter — Fri, 30 Jun 2006 07:27:06 GMT

You don't say whether this is a sudden change or perhaps a gradual decline or whether the slowness is in general or during a specific activity. Ruling out current system loading is the simplest thing to check and a good start.

If not sure whether this is a sudden change or not, if available, comparing the accounting information in the tail of log files of a batch job that runs regularly and puts roughly the same kind of load on the system (like backups) can indicate whether behavior is consistent over a long time.

Lots of places where a gradual degradation can occur, pool fragmentation, disk/file fragmentation, DB contention, resource contention.

If you suspect the network you may be able to devise a simple test to exercise it.

Happy Hunting,
Zeni

Re: VAx 7600

Robert Brooks_1 — Sat, 01 Jul 2006 20:33:14 GMT

What applications run on the cluster?

Is it only one of the three that is viewed as "slow"?

Is any attempt made to balance the load across the three systems?

-- Rob