HP GLANCE PLUS

[Andrew Luis Arruza]

Is there a way I can keep a history/log of the information that is charted out on the GlancePlus displays?
Thanks,
Andy

[John Palmer · Accepted solution]

MeasureWare agent process scopeux collects performance data that you can subsequently analyse by various tools.

It is a separately licenced product but is often bundled with glance.

Regards,

John

[Rita C Workman]

This is part of Measureware/Perf and you can build a 'history'. To do this requires a few things. You set up a file with your sar commands to tell it how often to 'gather' data on your system. You also need to set up your parameter file breaking down the different jobs and processes that you especially want more details on. The master/dummy is located at /var/opt/perf/parm. You can look over this file..and it will give you and idea of how to set things up. And then you have to run the MWA utility to make sure your parm file has no errors. This is a little too much to include in one small mail. You can review this information in the documentation for Glance (MWA and Perf). But I'd suggest (if you have support for this) talking to HP or see about taking the class on this.
Regards,

[Tim Malnati]

There is a good chance that Measureware is already on your system. The vast majority of people who purchase Glance buy the Plus Pack that includes Measureware. If your looking to chart this data out you will need Perfview also. This is separately a licensed product. It's also somewhat expensive, but your best bet if your looking to analyze any kind of performance metric over time. It is particularly handy to have if your trying to determine why you have been experiencing system degradation, for instance.

[Jim Welch]

You can't capture the gpm graphs themselves (not practically anyway), but you can gun glance in adviser mode to log any metric you can see in glance (and some others besides) to an ascii file that you can import into a spreadsheed and graph later. First, identify the metric name (like GBL_CPU_TOTAL_UTIL) using the built-in help or the metrics dictionary, then write and adviser script and run glance with the -adviser_only and -syntax switches and redirect the output to to a file. The online User's Guide should get you started.
Just as an example, say you want to log the info that gpm shows in the CPU Graph.
create a file (say /tmp/cpu.adviser) containing one long line:
print GBL_STATDATE," ",GBL_STATTIME," ",GBL_INTERVAL," ",GBL_RUN_QUEUE


Then run glance (not gpm) using the following syntax:
glance -j 30 -adviser_only
-syntax /tmp/cpu.adviser >> /tmp/adviser.out

This will give you the measurement date, time, interval (approximately 30 seconds due to the -j 30 switch), and the value of the metric GBL_RUN_QUEUE every 30 seconds until the glance process is killed. You can also use the -iterations switch to tell glance to make a certain number of measurements and exit.
NOTE: you can use any filename and either upper or lower case, I just use upper case for metrics to make the script easier to read.