AEN Version 4.x.x
This article addresses an issue when starting the wakari-compute service results in a FATAL ERROR. First you can check the log files to see the error resulting from starting the service. Log files are located at /opt/wakari-wakari-compute/var/log/wakari/, particularly the compute-launcher.application.log.
Error example:
10 Jan 15:34:49 - Deprecated: Use utils.get_pidfiles() instead FATAL ERROR: CALL_AND_RETRY_0 Allocation failed - process out of memory
This error does not necessarily mean there is a memory or disk space issue. If the above error is given it can be caused by the accumulation of .pid files left over from a possible shutdown of the service that resulted in a failure to cleanup these files. The compute service tries to load these all at once and exceeds the default memory allocation size of the service.
Solution:
You can delete all of the .pid files and directories located at /opt/wakari/wakari-compute/var/run/compute-launcher-apps/. These files are no longer needed as the processes they were attached to are no longer running. These files will again be created as new processes are started.