bug fixes and more

[bertsky]

• process_id must be a str not an int, esp. for process_mets.sh
• active job status must use PID from $REMOTEDIR/ocrd.pid (on Controller), not from $PID (on Manager)
• add missing job status D
• jobs: show (and sort by) time created/terminated
• use /workflows volume
• active jobs: add kill button (dummy for now)

[bertsky]

needs slub/ocrd_manager#57

[bertsky]

• add kill button (dummy for now)

@SvenMarcus to implement this, we would need to "log in" to the Manager and kill the given process ID. But I don't like the Monitor to hold the keys. Perhaps we should instead start implementing our minimalistic web server in the Manager itself – delegating to for_production.sh, for_presentation.sh (which we will need for new jobs anyway) and to the killer.

Sry, I'll update the CI tests shortly.

[markusweigelt]

Great work thx!

With the test, we are currently running on the CI in an endless loop.

tests/ocrdmonitor/test_readlogs.py .....                                 [ 37%]
tests/ocrdmonitor/test_redirect.py .........                             [ [53](https://github.com/slub/ocrd_monitor/actions/runs/4677162691/jobs/8284349346#step:5:54)%]
Error: The operation was canceled.

To me it looks like the tests/ocrdmonitor/test_sshps.py is affected.
Possibly it is due to the changed SSH user, since the test itself was not adjusted.

@SvenMarcus What do you think?

[markusweigelt]

All right thanks. When the pytest problem is solved I give my approve.

Everything except naming of ProcessState should be resolved