Collecting I/O trace - midaemon.

[Christof Schoeman]

Hi there

I want to collect an I/O trace and I believe that midaemon is the tool to use. However, I cannot find any documentation that tells me how to actually get to the information collected by midaemon (except for glance etc.).

Does anyone know how one would use midaemon for I/O trace collection?

Hope you can help.

[m saravanan]

You can get the I/O trace through Measureware. Basically "midaemon" is the process which gives data to MWA also. You can extract the data using "extract" command.
for details, pls see,

# man extract


Regds,
Saravanan

[Christof Schoeman]

I've looked at extract, but it seems as if scopeux interrogates midaemon every few seconds (configurable) and only writes summary information in the log files, where extract gets it from.

There also does not appear to be configurations options for scopeux to log each and every I/O in detail.

Unless I'm missing something.

[Tiziano Contorno _]

Hi Christof, have a look at:

/opt/perf/examples/mwaconfig/parm_apps
/var/opt/perf/parm #to create groups and tune thereshold
man mwa
man midaemon
Extractable data have a "granularity" of 5 min :(
Default is 60 min.
Value reported is the media of values in the interval.
Mind that if the thereshold is not reached the value extracted is as set in reptfile MISSING value.

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=1006501


Regards.

[Bill Hassell]

midaemon is not accessed directly -- it's purpose is to very quickly collect hundreds of measurement values into a shared memory area where Measureware (man mwa) reads the data and stores it in the mwa log files. This must take place extremely fast as the data changes hundreds to thousands of times per second. The extract command can pull specific metrics from the logfiles at any time. The extract options allow creating a file which can be imported into a spreadsheet program such as Excel. A list of the metrics is found in the Measureware docs directory (/opt/perf). The logs are configured to collect a fixed period of time to prevent massively large log files.