Closed KshitijT closed 4 years ago
It did not do this before...
I will test this and hopefully not reproduce. Has the implementation of mstransform in stimela changed during this week (e.g. were we using 4.something before and not now?)
I don't think so; please see if it is reproducible. I don't know why I did not encounter this earlier - maybe I just used the minimal or meerkat configs.
I will also look into it today.
@KshitijT @paoloserra I started this yesterday fresh from master. Had to edit carateConfig.yml for keywords clean_mask_threshold: [20,10,5,5] clean_threshold: [0.5,0.5,0.5,0.5] in selfcal worker. I am not finished yet because I want to let it run through and see if I can confirm a report of stripes reappearing, but what I got so far is:
Starting container [modelsub_0-14029794096062415859569100829802]. Timeout set to -1. The container ID is printed below.
Running: docker start -a modelsub_0-14029794096062415859569100829802
Successful read/write open of default-locked table /home/jozsa/msdir/1524929477-circinus_p1_l.ms: 26 columns, 47616 rows
Wrtiting to column CORRECTED_DATA (rows 0 to 4760)
Wrtiting to column CORRECTED_DATA (rows 4761 to 9521)
Wrtiting to column CORRECTED_DATA (rows 9522 to 14282)
Wrtiting to column CORRECTED_DATA (rows 14283 to 19043)
Wrtiting to column CORRECTED_DATA (rows 19044 to 23804)
Wrtiting to column CORRECTED_DATA (rows 23805 to 28565)
Wrtiting to column CORRECTED_DATA (rows 28566 to 33326)
Wrtiting to column CORRECTED_DATA (rows 33327 to 38087)
Wrtiting to column CORRECTED_DATA (rows 38088 to 42848)
Wrtiting to column CORRECTED_DATA (rows 42849 to 47609)
Wrtiting to column CORRECTED_DATA (rows 47610 to 47615)
Container [modelsub_0-14029794096062415859569100829802] has executed successfully
Runtime was 0:00:02.083625.
Will give full report when finished. Should be soon.
It definitely went through, now mosaicing.
@KshitijT Is this happening with docker or singularity? I tested with Docker.
Docker.
On Sat, Apr 4, 2020, 11:16 Gyula Jozsa (Josh) notifications@github.com wrote:
@KshitijT https://github.com/KshitijT Is this happening with docker or singularity? I tested with Docker.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ska-sa/meerkathi/issues/891#issuecomment-609000772, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACGSY6U2LR5UQZOJSMSKXBTRK33G7ANCNFSM4L4KPADA .
Hold on, I see what happened.
While testing #888 , I turned off transfer_model, which means self_cal was done on MSs with label "cont" but the model was never passed on to the ones with label "corr" (which did have the corrected data column though, which was duly generated by transfer_apply_gains.
...Which is why the modelsub complains in image_line. Apologies @gigjozsa and @paoloserra , for wasting your time.
Whatever happened to the emoji for "feeling stupid" ?
It does so with this message:
The relevant bit of carateconfig is :
I suspect this happens because even with
column: 'all'
, mstransform does not copy over theMODEL_DATA
column, modelsub tries to subtract the nonexistentMODEL_DATA
column from theCORRECTED_DATA
and crashes. See:Of course this does not happen with `stanard' usage of the pipeline, since there is no intervening split step between selfcal and line_image workers.
Has anyone else encountered the same error?
The solutions, if the code is to be changed, might be:
So I don't know if this constitutes a bug. Up to @paoloserra and @gigjozsa if to amend Caratekit config or to change the code.