[HTCondor-users] Using authentication

[Hugh Jennings <hugh@xxxxxxxxxxx>]

Hi,

I have a simple setup of 2 linux machines.

If I submitted a job from the system running  schedd, the collector and the negotiator  it was possible to submit jobs and have them run on all available slots.  If I tried to submit the same job from the system only running the MASTER, PROCD and STARTD only the jobs would be accepted and immediately put on hold.  condor_q -analyze indicated there was an authentication problem.

Searching the local and general config files did not give me a clear idea of where the authentication methods were being set.   From the web documentation I tried to setup simple PASSWORD authentication. 

No errors were apparent when I created the password file using condor_store_cred -f /etc/condor/pool or condor_store_cred -c add.

Unfortunately now the cluster is broken in the sense that no jobs get accepted regardless of what node I send it from.

The error message I get is:

06/20/13 00:10:43 (pid:18245) DC_AUTHENTICATE: authentication of <192.168.1.183:45992> did not result in a valid mapped user name, which is required for this command (1112 QMGMT_WRITE_CMD), so aborting.

Googling this is not giving me any help.

The lines I added to condor_config are:

SEC_DEFAULT_AUTHENTICATION_METHODS = PASSWORD
SEC_PASSWORD_FILE = /etc/condor/pool
SEC_DAEMON_AUTHENTICATION_METHODS = PASSWORD
SEC_CLIENT_AUTHENTICATION_METHODS = PASSWORD
ALLOW_DAEMON = condor_pool@$(UID_DOMAIN)/*.my.domain, \
               condor@$(UID_DOMAIN)/$(IP_ADDRESS)
ALLOW_NEGOTIATOR = condor_pool@$(UID_DOMAIN)/oracle1.my.domain
ALLOW_ADVERTISE_STARTD = condor_pool@$(UID_DOMAIN)/*.my.domain

Any help would be appreciated

Hugh