[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Condor-users] Scheduling problem


I'm trying to build a condor-based cluster where the first node will be
used to submit jobs and the rest of the nodes will be exclusively used
to run the different jobs submitted. I mean, the users have to submit
all the jobs from this node and they can't connect to the rest of the
nodes (i.e one shared queue for all the nodes). But now, when I submit
some jobs from one node, these jobs are just scheduling in this node,
remaining the rest of the nodes as "unclaimed". 

During the initialization process I get these error messages in the
Collector logfile:

9/26 21:10:29 In CollectorDaemon::Config()
9/26 21:10:34 enable: Creating stats hash table
9/26 21:10:54 (Sent 0 ads in response to query)
9/26 21:10:54 WARNING:  No master ad for < node01.cluster >
9/26 21:10:54 ScheddAd     : Inserting ** "< node01.cluster , >"
9/26 21:10:54 stats: Inserting new hashent for
9/26 21:10:54 condor_write(): Socket closed when trying to write buffer
9/26 21:10:54 Buf::write(): condor_write() failed
9/26 21:10:54 SECMAN: Error sending response classad!

As far as I know I have no firewalls running in the machines.

Could anyone give me a clue?

Thanks in advance.

Best regards