Distributed Computing Server (2015a) network issue

I'm submitting a job from a client computer.

I get the following error on the last step:

VALIDATION DETAILS
Profile: gaia
Scheduler Type: MJS

Stage: Cluster connection test (parcluster)
Status: Passed
Description:Validation Passed
Command Line Output:(none)
Error Report:(none)
Debug Log:(none)

Stage: Job test (createJob)
Status: Passed
Description:Validation Passed
Command Line Output:(none)
Error Report:(none)
Debug Log:(none)

Stage: SPMD job test (createCommunicatingJob)
Status: Passed
Description:Validation Passed
Command Line Output:(none)
Error Report:(none)
Debug Log:(none)

Stage: Pool job test (createCommunicatingJob)
Status: Passed
Description:Validation Passed
Command Line Output:(none)
Error Report:(none)
Debug Log:(none)

Stage: Parallel pool test (parpool)
Status: Failed
Description:The validation stage encountered a MATLAB exception.
Command Line Output:(none)
Error Report:
Failed to initialize the interactive session.

Caused by:
    Error using parallel.internal.pool.InteractiveClient>iThrowIfBadParallelJobStatus (line 765)
    The interactive communicating job errored with the following message: No peer connected to MatlabPoolPeerInstance{fUuid=4c8d3b80-4e34-4ee3-ace6-71b363457cbb, fGroupUuid=ad507a2c-4416-4b80-89cd-23846719e5dc, fLabIndex=33, fNumberOfLabs=64} on compute-18/192.168.2.118:14384 after polling accept for 60000 milliseconds using gaia/192.168.2.125:14351: Timed out after 60000 MILLISECONDS
Debug Log:(none)

In admin centre, all hosts are up and communicating. MJS is running. I have 64 worker nodes running (all reporting green status).

I'm using a Torque/PBS system.

[Edric Ellis]

I would suggest contacting MathWorks technical support to help resolve this installation problem.