When job creation takes to long, ecflow server writes a message to the log file. It can be slow for a variety of reasons.
- The disk where the server is running is slow.
- The number of includes in the .ecf may be excessive, causing very large job files.
- Preprocessing takes to long. (i.e. replacing %VAR% in the .ecf file, with variables in the Node tree), removing comments and manual sections, etc.
- The server may be running out memory.
- The server is starved of CPU time, competing with other processes.
When the ecflow server is started, the ECF_TASK_THRESHOLD is read from the environment.
Job creation taking longer than this threshold is written as a warning in the log file. The default threshold is 4000 milliseconds seconds.
However, in most cases this was ignored by the user until some other problem arises.
In ecflow 5 series, they will be notified in the GUI that the task has exceeded this threshold, by a new icon. This will allow early diagnoses of a looming problem.