Closed cyriltasse closed 1 year ago
when you say you've fixed the no_beamcor
what do you mean?
@jwpetley just pushed a fix for this problem.
restarting the generic pipeline from a specific step requires editing the statefile - there is a prefactor tool or if you're old school then you can edit it directly, but that requires being familiar with the statefile. If it's failed on the delay_solve step then you'll probably also have to delete the delay_solve
directory in the working directory, if it got as far as creating that; this goes for any step that fails or you want to re-do -- if the output already exists, the pipeline will fail (it won't fail if the step is successful in the statefile).
If the Agg/TkAgg problem is coming from facetselfcal.py then it's best to open an issue there: https://github.com/rvweeren/lofar_facet_selfcal
They removed the no_beamcor
argument from the facetselfcal.py script entirely. There is now instead an argument beamcor
which can be yes
, no
or auto
. I pushed the change so that the lofar-vlbi pipeline is still compatible with the latest version of the selfcal script. It would crash as above otherwise. I think @cyriltasse has seen and done the same fix.
I'm unsure what is going on with the Agg issue. I haven't seen that before. I ran a full run of facetselfcal.py yesterday using the lofar-vlbi pipeline yesterday without issue.
They removed the no_beamcor argument from the facetselfcal.py script entirely. There is now instead an argument beamcor which can be yes, no or auto. I pushed the change so that the lofar-vlbi pipeline is still compatible with the latest version of the selfcal script. It would crash as above otherwise. I think @cyriltasse has seen and done the same fix.
Yes I had seen your fix @jwpetley and changed the code in both the [...]/Delay-Calibration/delay_solve/
and pulling the latest change in lofar-vlbi/master. What's weird is that the pipeline crashes, but the no_beamcor
error is quoted as a warning... Actually there are no ERROR
errors, apart from the final
2023-04-14 12:27:17 ERROR genericpipeline: *******************************************
2023-04-14 12:27:17 ERROR genericpipeline: Failed pipeline run: Delay-Calibration
2023-04-14 12:27:17 ERROR genericpipeline: Detailed exception information:
2023-04-14 12:27:17 ERROR genericpipeline: <type 'exceptions.KeyError'>
2023-04-14 12:27:17 ERROR genericpipeline: 'mapfile'
2023-04-14 12:27:17 ERROR genericpipeline: *******************************************
restarting the generic pipeline from a specific step requires editing the statefile - there is a prefactor tool or if you're old school then you can edit it directly, but that requires being familiar with the statefile. If it's failed on the delay_solve step then you'll probably also have to delete the delay_solve directory in the working directory, if it got as far as creating that; this goes for any step that fails or you want to re-do -- if the output already exists, the pipeline will fail (it won't fail if the step is successful in the statefile).
Ok - we're getting to close to the week-end, so I'll restart from scratch the delay calibration. Next time I'll try to dig a bit into how things are wired, and restart from last failed step.
Thanks al for all the info. I close this for now, I'll reopen if it crashes after that no_beam fix
Hi,
I'm not sure how to debug this: the pipeline
Delay-Calibration.parset
finished with the fillowingI don't see any real errors - just warnings. I'm fixed the
no_beamcor
, and not sure if there's an easy solution to the Agg/TkAgg one. Where should I go from there?A side question: with the generic pipeline, how do I start over from a given pipeline step?