Xavier,
Thanks for the note.
--- You wrote:
I've discovered the new IM probe: InterMapper Engine Status.
It shows some technical informations and the efficiency of the engine but....
Could you tell us how that's done and for what value we can say that the server
is not fast enough ?
What values are important and how could we interpret them ?
--- end of quote ---
The InterMapper Engine Status probe monitors the state of InterMapper's polling
process. We created it in the course of debugging the high-packet loss problem
that certain customers reported.
The probe makes a number of measurements that are described in the Set Probe...
window. There's an updated description in the probe of version 5.0.7b3 whichI
have appended for your information.
The theoretical maximum polling rate is 66.66 Hz (every 15 msec). The probe also
displays the actual rate as a percentage of the maximum (this was called "Efficiency"
in earlier versions of the probe, but that name was misleading). As long as the
main loop frequency (in Hz) is above 30, everything will be fine. We have seen
high packet loss when the frequency gets near or below 10 Hz. The other numbers
are primarily displayed to supply information to Dartware engineers.
Please get back to me if you have further questions. Thanks.
Rich Brown [email protected]
Dartware, LLC http://www.dartware.com
66-7 Benning Street Telephone: 603-643-9600
West Lebanon, NH 03784-3407 Fax: 603-643-2289
===== Description from InterMapper Engine Status probe =====
InterMapper Engine Status
This probe monitors the status of the InterMapper polling engine. With the default
setting, this probe displays the results of 500 loops through the polling engine.
To measure activity at a finer-grain, decrease the value of the loops parameter.
Setting the "loops" parameter to the value 1 will update the statistics on every
pass through the main run loop.
The "Main Loop" frequency is the number of times that InterMapper performs the
main loop each second. The theoretical maximum loop frequency is 66.667 loops
per second, based on the current yield value of 15 msec. If it falls below 10
or even 5 loops per second, InterMapper may report false outages.
This probe also reports polling rate as a percentage of the maximum loops per
second. This is a measure of how much additional processing occurs per loop. This
percentage will never be 100%. It should, however, level out and remain steady
over time.
On Unix systems, this probe reports Context Switches Per Loop (CSPL). This is
another measure of the overhead of InterMapper's processing as it runs on your
system. Fewer context switches per loop is better (ideal = 0), since context switches
carry overhead. A server with thousands of devices and hundreds of mays may well
have a CSPL greater than 2 during normal operation. (This value is not available
on Windows systems, and is alway set to -1.)
InterMapper tracks the number of bytes sent out the main UDP polling socket. Bytes/Loop
is the average bytes sent per loop, averaged over the last batch of N loops. Bytes
Peak is the maximum number of bytes sent in a *single* polling loop. (In the
current implementation, the peak bytes is checked on every loop, but only resets
to 0 when you change the # loops parameter; ie peak bytes is not the peak bytes
of the last batch of N loops.)
____________________________________________________________________
List archives:
http://www.mail-archive.com/intermapper-talk%40list.dartware.com/
To unsubscribe: send email to: [email protected]
