topic Re: system statistics in Operating System - OpenVMS

system statistics

mustafa_12 — Mon, 02 Jan 2006 17:04:42 GMT

Hi all,

I want to use some monitoring commands in DCL scripts in order to determine CPU usage, disk I/O rate, network interface states (utilization, failure, etc...), memory usage, etc. I put these commands in a .com file and run it in regular interval. According some threshold values, the the program gives an appropriate alert.

Can I use the monitor command like "monitor cluster /summary=out.txt"? But, with this command, it records the statistics to the output file continously. I can use /ending qualifier but I can not determine how much time it takes to take full enough statistics. I mean, the output file should contain exactly one statistics per item, not two or more. One CPU usage per cluster node, one I/O rate per disk, one memory usage per cluster node, one network statistics per card, etc... I do not want second statistic for the same item, two CPU statistics per cluster node for instance.
Is it possible, with monitor or other commands..

thanks...

Re: system statistics

Robert Gezelter — Mon, 02 Jan 2006 23:14:11 GMT

Mustafa,

Check out the T4 add-on, available from the OpenVMS www site. There is also an article on T4 (see http://h71000.www7.hp.com/openvms/journal/v3/index.html#t4)

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

Re: system statistics

Wim Van den Wyngaert — Tue, 03 Jan 2006 01:40:21 GMT

I use monitor in dcl to get some info.
See attached file.

Wim

Re: system statistics

Andy Bustamante — Tue, 03 Jan 2006 12:17:47 GMT

I'll second Bob's endorsement of T4. You can get the kit at http://h71000.www7.hp.com/openvms/products/t4/index.html TLViz is a PC utility to create graphs from T4 data files.

You may also want to look at the ECP Data Collector and Performance Analyzer. See http://h71000.www7.hp.com/openvms/products/ecp/performance-and-capacity-download.html

Re: system statistics

Paul Jerrom — Thu, 05 Jan 2006 15:15:22 GMT

Howdy,

I do something similar to Wim. Attached is a modified piece of DCL from a system checker I use. This part checks there isn't a cpu hog (defined by me as a process using 85%+ cpu 15 minutes apart, which catches most of my runaway processes). Shows the principal of using a monitor file, anyway, so can be adapted to meet your needs. BTW, works on VMS Alpha V7.3-1, other versions/platforms may have different number of records in the output file, different location of data within the record etc.

Have fun,

PJ