Mailing List ArchivesAuthenticated access |
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}
|
Mailing List ArchivesAuthenticated access |
|
|
Hello. On the Windows platform, we configure all machines in a pool for PASSWORD authentication. This requires adding the same pool password on all machines using âcondor_ctore_cred add âcâ. That works fine, it seems it only requires that the
condor_master daemon be running. We would like to provide functionality that indicates the validity of the pool password. In principle, that should be possible using âcondor_store_cred query âcâ. However, this has a couple of limitations: Â
It only indicates whether a pool password is stored locally. It does not indicate whether it matches the password on the pool manager (or the machine identified by CREDD_HOST). So it might report that the credentials are valid
when in fact the password is incorrect. Â
It only works if the machine is running the condor_schedd daemon. This begs a couple of questions: 1.
Is there a good reason why querying the pool password existence requires the condor_schedd whereas adding or deleting it does not? Since the password is required on all processing nodes, it seems odd that its existence cannot easily
be queried on those nodes. 2.
Is there some other command that can be run to test the validity of the pool password with respect to the CREDD_HOST? On a processing node, one can wait for the condor_startd to fire up and then inspect its LocalCredd class ad, but it
would be good if there were some more direct means, e.g. some variation of the âcondor_statusâ command? Thanks, David Wilkins
|