Re: [Condor-users] Central Manager performance during matchmaking

[Ian Chesal <ICHESAL@xxxxxxxxxx>]

> I would appreciate if you can share with me the example of condor
classAd with simple
> description vs. complex descriptions. I will use the given classAds to
submit to the
> CM and monitor the workload of the condor_negotiator daemon. This
experiment is
> actually to study the performance/scalibility of my CM based on the
increasing number
> or the complexity of the classAds.

You actually have 4 cases to measure:

Case 1: Simple job ad, simple slot ad
Case 2: Simple job ad, complicated slot ad
Case 3: Complicated job ad, simple slot ad
Case 4: Complicated job ad, complicated slot ad

And even that is...uh...simplifying the measurement cases considerably.

For Case 1: Create a slot that has no restrictions on. Create jobs that
have no "Requirements =" lines in them that you know will match to the
slot.

For Case 2: Make the Start expression and/or the Rank expression on the
slot something funky with time-of-day logic and username logic. Use the
same jobs as in Case 1.

For Case 3: Add a Requriments string to your job ad that looks at only
Condor-default attributes on a machine. Use the same machine ad as in
Case 1.

For Case 4: Same job ads as in Case 3 but now add in the machine ads
from Case 2. Maybe define custom attributes on the machine and match
against those in the jobs.

Sky's the limit on how "complicated" you want to get with the
complicated case.

- Ian

Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution,  or copying  of this message, or any attachments, is strictly prohibited.  If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments.  Thank you.