Re: [HTCondor-users] Setup of flocking

[John M Knoeller <johnkn@xxxxxxxxxxx>]

If your primary pool was 9.12 or later, I would suggest that you use IDTOKEN authentication between the two pools.  But 8.6 does not support IDTOKEN.

 

I don’t think that trying to disable authentication is going to work for flocking.   This is something we never do, and so we don’t have any suggestions.

 

That leaves you with using either SSL or PASSWORD authentication between the two pools.    Of the two options PASSWORD is easier to set up.  You just need to set a common pool password between the SCHEDD and the remote central manager using

 

    condor_store_cred -c

 

And make sure that PASSWORD authentication is in the list of authentication methods (it is by default).

 

There are basic instructions for enabling PASSWORD authentication here https://htcondor-wiki.cs.wisc.edu/index.cgi/wiki?p=HowToEnablePoolPassword

 

The various SEC_* configuration variables should already have the right defaults so you probably don’t need to modify your configuration unless you have set the SEC_* variables already.

 

Note that the default value for SEC_PASSWORD_FILE is different between 8.6 and 9.12, but since these will be files on different machines, that does not matter,   The contents of these two files will need to be the same though,  and the file content should not contain any NUL characters, since 8.6 does not handle them correctly.

 

-tj

 

From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> On Behalf Of Peter Ellevseth
Sent: Monday, February 13, 2023 8:02 AM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
Subject: [HTCondor-users] Setup of flocking

 

Hi all

 

I am having some issues with setting up flocking on our cluster. We are working with an external supplier to provide some additional computing power in times of need. Our intent was then to establish a separate condor cluster on their infrastructure and use flocking to utilize the additional power. However, I am not able to get the two pools to talk together. I set up the macros on one of our internal schedd-machines and in the general config-files on the external pool (in the config.d folder) as specified here:

https://htcondor.readthedocs.io/en/latest/grid-computing/connecting-pools-with-flocking.html

 

After a lot of back and forth, it is still not working. Initially I got into trouble with authentication, getting error messages like ‘SEMAN: FAILED: Received DENIED:…’

 

In an attempt just to get it running I tried disabling authentication by setting relevant SEC_* macros to NEVER and ANONYMOUS. I am still being told that the two machines are unable to authenticate eachother. Our internal pool is running condor 8.6, while the external instance is running 9.12. Could that be an issue?

 

What would be the preferred method of authentication for this kind of setup? I feel the manual was somewhat short on what is necessary to set up flocking. Is there any supplemental information anywhere?

 

Any tips would be great.

 

Peter

 

 

Peter Ellevseth
Principal Advisor / Principal Advisor
	+47 93 43 56 01 / +47 73 90 05 00
	peter.ellevseth@xxxxxxxxxx
	safetec.no