Logger names must be descriptive

[gijzelaerr]

Most of the Trap logger names reflect simply the name of the source file. For example, I have a log like the following:

2013-03-11 14:30:10 INFO    trap.run.distributed: Running task: null_detections
2013-03-11 14:30:10 INFO    trap.run.distributed.null_detections: recipe null_detections started
2013-03-11 14:30:10 INFO    trap.run.distributed.null_detections: starting null_detections for images [2632]
2013-03-11 14:30:10 INFO    trap.run.distributed.null_detections: Limiting to 8 simultaneous jobs/node
2013-03-11 14:30:10 INFO    trap.run.distributed.null_detections: Waiting for compute threads...
2013-03-11 14:30:12 INFO    trap.ingredients.source_extraction: Forced fitting in image: /home/antoniar/RSM/run0/L89691_SAP002_band1.img.restored.corr
2013-03-11 14:30:15 INFO    tkp.database.utils.general: Inserted 6 forced-fit null detections in extractedsource for image 2632
2013-03-11 14:30:15 INFO    trap.run.distributed.null_detections: recipe null_detections completed

In the single-threaded case we see here, it's clear what's happening. However, this doesn't scale: when there are multiple null_detections scripts running across several compute nodes, there will be no way to tell which message is from which script. The logger should note (at least) the node it's running on and (preferably) the file it's processing.

Gijs Molenaar: true. this can probably be set by configuring the logger in a better way. It supports a lot of features, like line number and process ID. probably it can also be massaged to accept something like a hostname.

moved from LOFAR issue tracker
https://support.astron.nl/lofar_issuetracker/issues/4289

[AntoniaR]

Unclear if this is still an issue so would be good to check with the latest version.