I recently added a Linux machine which has 24 slots (and hence can run 24 jobs) to our condor pool.
After around 8 jobs started running on the Linux machine, Central Manager could not send any more jobs, and jobs are failing with Shadow Exception.
Can no longer talk to condor_starter <xxx.xxx.xxx.xxx:9604>
Error from slot8@xxxxxxxxxxxxxxxx: Could not initiate file transfer
Linux machine is behind the firewall and we have opened the firewall for the below port range.
Looks like the condor_starter which is using one of the ports from the above range, in this case port 9604, could not handle more than 8 connections and hence could not run more than 8 jobs seems.
Can anyone suggest how this can be fixed so that jobs can be run on all the 24 slots.