Solved

Child process is stuck on queue when using FMEServerJobSubmitter

  • 21 August 2023
  • 5 replies
  • 38 views

Badge +4

I have a batch workspace that spawns multiple child workspaces using four separate FMEServerJobsubmitters. I have two engines in FME FLow but the other engine is occupied with other processes so I'm able to use only one engine.

 

As i understood, if i set Wait for jobs to complete to "Yes" and Submit Jobs "In Sequence", FME Flow should spawn a sub-engine that will run the child workflows and once they are ready it will continue the parent process. This is the behaviour that I would need because I only want to continue parent workflow when all the child processes are successful. However, with this setup the first child process enters queue and never starts. It also shows in the parent log file that the submit was successful, but it ends up into queue.

 

I have separate queues for the two engines and both the parent process and the child process are tied to engine2 running in queue2 so it should interfere with the other engine and queue. However, it seems that the parent workspaceis blocking the child workspace from running.

 

I Tried to troubleshoot with these https://community.safe.com/s/article/fme-server-troublshooting-fmeserverjobsubmitter#QUEUED

instructions but didn't find anything i could check anymore. Have I understood

the FMEServerJobSubmitter parameters wrong somehow?

icon

Best answer by jnknnn 28 March 2024, 14:57

View original

5 replies

Userlevel 2
Badge +17

Hi @jnknnn​,

If you have the option Submission Mode - Submit Jobs in the FMEServerJobSubmitter, please try setting it to In Sequence. According to the documentation, this should allow the workspace to proceed even though only one engine is available.

Badge +4

Hi @jnknnn​,

If you have the option Submission Mode - Submit Jobs in the FMEServerJobSubmitter, please try setting it to In Sequence. According to the documentation, this should allow the workspace to proceed even though only one engine is available.

I have set submission mode to In Sequence

image

Userlevel 2
Badge +17

I have set submission mode to In Sequence

image

I did a quick test on FME Flow 2023 with only one engine activated, and a workspace with four FMEFlowJobSubmitters worked OK, both in parallel and in series.

However, we have had problems in past versions of FME Server if the child workspace is set to use a different Python version than the parent, or if the child workspace crashes during its run.

 

What version of FME Server are you using?

 

Would you mind sharing the workspace with the FMEServerJobSubmitter in it, and the log file from its run?

Badge +4

I have set submission mode to In Sequence

image

We are using Fme server  2021.2.4. and workspaces are made in desktop 2021.2.1.0. I checked that all workspaces should be using python 3.8+. Parent workspace is attached, I was not able to access log-file anymore. However, the last line in the log file was "FMEServerJobSubmitter_3 (ServerFactory): <fmeserverhost>- Running workspace '<workspacename>' in repository '<repositoryname>'... and the first child workspace entered queue.

 

I removed sql's, connection parameters and other connection information for ip and other obvious reasons but other than that it should work as intended. In all FMEServerJobSubmitter transformers i had connection parameters embedded.

Badge +4

I was able to solve this by using default queue in FMEServerJobSubmitter and not assigning engine to the default queue. For some reason using any custom queue assigned for specific engine didn’t work. 

Reply