Subject: | |
From: | |
Reply To: | |
Date: | Thu, 6 Mar 2008 22:06:55 +0100 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Paul,
> When trying to 'drill down' to the process level using
> Perfview , I only receive one day's worth of ( process ) data, though
> CPU Util will show me the whole month ...any ideas
the process detail metrics typically use much more space than
global metrics, so for a given log file size there is much more
history for the global metrics.
You can use the SCAN command in UTILITY.SCOPE to get info about
the sizes of the different logfiles (logglob, logappl, logproc,
etc) and how much records and megabyte per day they collect.
You can also use UTILITY.SCOPE to resize some or all of your log
files accordingly, if you want to have a week's worth of process
details, for example.
As the scope logs are "circular" in nature, i.e. when getting full
they throw away the oldest data to make room for new data, you may
want to also implement an archiving strategy with EXTRACT.SCOPE in
addition. This will allow long-term access to the metrics.
Chapter 4 and 5 in the Scope/PCS manual discuss the UTILITY and
EXTRACT tools. Chapter 6 has a couple of tips for log archiving.
Also see http://docs.hp.com/cgi-bin/doc3k/B5070090038.12050/1
Lars.
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|
|
|