PP 8.5_CU3 on an HPC environment - Crash when using LSF

When installing and running Pipeline Pilot 8.5_CU3 on an HPC Cluster (LSF queuing system v7.06) with RHEL 6.2 – 64 bit and PP installed on a shared storage partition (Lustre/SFS), then running any task results in the following error messages below.  Note that MS601 runs well on this platform. Also, tasks run well if the queuing system is not used (simply launch jobs on the head node).

Error message:

Server Error Report: Pipeline Pilot error caught while running the protocol: connect failed in tcp_connect()

SOAP Faultcode: SOAP-ENV:Client. Error code: 24 SOAP Faultstring: Connection refused

soap_call_sci__GetServerConfig() failed

CLocatorSoap::GetServerConfig: Pipeline Pilot exception rethrown

iLocatorSoapImpl::GetServerConfig: Pipeline Pilot exception rethrown

Server Error Report: Pipeline Pilot error caught while running the protocol: connect failed in tcp_connect() SOAP Faultcode: SOAP-ENV:Client. Error code: 24 SOAP Faultstring: Connection refused

soap_call_sci__GetServerConfig() failed

CLocatorSoap::GetServerConfig: Pipeline Pilot exception rethrown

iLocatorSoapImpl::GetServerConfig: Pipeline Pilot exception rethrown

Configuration info (Administration portal):

  1. Setup -> Server
    • Maximum no cpu’s to use set to the number of cores on HPC.
  2. Setup->Clustering
      • Load balancing off (does not change situation if this is changed to Load Levelling).
      • Private Cluster ticked (since access to/from compute nodes is only via head node)
    • Setup->Grid
          • Grid engine type: LSF
          • Path: /lsf (lsf.conf is located under /lsf/conf/lsf.conf)
          • Run on grid by default – On (then errors occur).

        Any assistance would be appreciated.

        Regards

        Enrico