Author Topic: Qube 6.5 Submitting test job goes to pending and all workers disconnect  (Read 11278 times)

Rayminette

  • Jr. Member
  • **
  • Posts: 7
Hello,
   I'm trying to submit Qube test (set) job.  When I do, it immediately goes to a pending state and all the workers are suddenly listed as down and they stay that way until I restart the worker service on the workers.
   Why does submitting a job immediately disconnect all the workers? How can I fix this?

Thanks

jburk

  • Administrator
  • *****
  • Posts: 493
Re: Qube 6.5 Submitting test job goes to pending and all workers disconnect
« Reply #1 on: August 14, 2014, 07:58:52 PM »
I'll bet that the firewalls are running on your workers.

The workers can talk to the supe through their firewalls, so the supe sees them as up.

But as soon as the supervisor tries to dispatch a job to the workers, it can't reach them due to the workers' firewalls blocking the supervisor, so the supervisor then thinks that the workers are down.

Rayminette

  • Jr. Member
  • **
  • Posts: 7
Re: Qube 6.5 Submitting test job goes to pending and all workers disconnect
« Reply #2 on: August 15, 2014, 01:56:01 PM »
I'll bet that the firewalls are running on your workers.

The workers can talk to the supe through their firewalls, so the supe sees them as up.

But as soon as the supervisor tries to dispatch a job to the workers, it can't reach them due to the workers' firewalls blocking the supervisor, so the supervisor then thinks that the workers are down.

Nope, all firewalls are off.  It had been working fine until we tried to update to 6.6.1.  We had so many errors with it, that we rolled back to 6.5. 
And this is the only thing that is persisting now.

Workers show as idle in the list.  Submit test job, it immediately goes into a pending state.  Worker stays as down until the job is killed and the service is restarted.