client side test frequency tuning 🔗 link

Hello all,

 
Once again a little question about frequency changes in the hobbit tests:

The CPU, disks, procs kind of tests are held on the client side right? What
I want to do is to change the frequency of the CPU survey without changing
anything for the other tests. Since the client side tests are launched all
together by hobbitclient.sh which is programmed to launch them at the
frequency defined in clientlaunch.cfg, there is no "easy" way to set
different frequencies.

 
To me it could be possible to set them up by modifying clientlaunch.cfg to
put new entries specifying the new frequencies. The entries would launch
modified hobbitclient.sh scripts so that they would launched the good OS and
command line specific scripts. But what are exactly the consequences on the
data fetching? I have read somewhere that it could have consequences and my
question is:

 
Is there a workaround?

 
Etienne Marganne,

TI Automotive.


The information contained in this transmission may contain privileged and confidential information.  It is intended only for the use of the person(s) named above. If you are not the intended recipient, you are hereby notified that any review, dissemination, distribution or duplication of this communication is strictly prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.

On Thu, Jan 25, 2007 at 03:03:20PM +0100, Marganne, Etienne wrote:

No. That's Big Brother thinking, and you're running Hobbit.

There is no one-to-one correspondance between the client-side script and the cpu/disk/memory etc. tests that shows up on the Hobbit display, since the latter is generated on the Hobbit server from several of the data items that the client-side script collects.

What I want to do is to change the frequency of the CPU survey without changing anything for the other tests.

You can modify the clientlaunch.cfg to run the full Hobbit client script
every minute, but that would probably break your log-monitoring (log
events would expire in 6 minutes instead of 30), and the vmstat data
would probably be affected as well.


Besides which, it doesn't really make much sense. The CPU load reported
is the 5-minute average from your OS anyway. And do you really want a
sudden spike to trigger full alerts? Just because someone happened to
run "find" in the wrong directory?


Regards,
Henrik

First the CPU test was chosen as an example it sure does not make much sense
to accelerate the frequency of the data collection for that ;-).

About the main topic, I think I have been confusing people a lot with term
misuses. What I would like to do is to accelerate or slow down the frequency
of data collection, not the tests that are held, as you said, on the server
side.
The idea is to try to tune the data collection frequencies the most
accurately possible.

You said that I can change the frequency of the execution of the full Hobbit
client... That is what I do not want to do. But you also said that it would
break up the log file ... only because rising the data collection flow would
be really flood like (thing that I strongly believe possible) or is there
another reason?

Regards,
Etienne Marganne,
TI Automotive.

-----Original Message-----
From: user-ce4a2c883f75@xymon.invalid [mailto:user-ce4a2c883f75@xymon.invalid] 
Sent: jeudi 25 janvier 2007 16:04
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] client side test frequency tuning

On Thu, Jan 25, 2007 at 03:03:20PM +0100, Marganne, Etienne wrote:

No. That's Big Brother thinking, and you're running Hobbit.

There is no one-to-one correspondance between the client-side script and 
the cpu/disk/memory etc. tests that shows up on the Hobbit display, since 
the latter is generated on the Hobbit server from several of the data items 
that the client-side script collects.

You can modify the clientlaunch.cfg to run the full Hobbit client script
every minute, but that would probably break your log-monitoring (log
events would expire in 6 minutes instead of 30), and the vmstat data
would probably be affected as well.


Besides which, it doesn't really make much sense. The CPU load reported
is the 5-minute average from your OS anyway. And do you really want a
sudden spike to trigger full alerts? Just because someone happened to
run "find" in the wrong directory?


Regards,
Henrik


The information contained in this transmission may contain privileged and confidential information.  It is intended only for the use of the person(s) named above. If you are not the intended recipient, you are hereby notified that any review, dissemination, distribution or duplication of this communication is strictly prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.