Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: drp_pipe
-
Labels:None
-
Story Points:6
-
Epic Link:
-
Team:Data Release Production
-
Urgent?:No
Description
Lee Kelvin and I encountered an unusual error while attempting to run some DECam (Merian) data through step2d on the Tiger cluster at Princeton. The error given is this:
lsst.pipe.base._status.InvalidQuantumError: No 'input_exposures' with detector 25 for visit 1068555 even though this detector is present in the input visit summary catalog. This is most likely to occur when the QuantumGraph that includes this task was incorrectly generated with an explicit or implicit (from datasets) tract constraint.
|
We use BPS/parsl to submit this job on w_2023_10, and the data query for our step2d only contains visit=. When we initialize a run, the finalizeCharacterization task appears to complete successfully, but the second task, updateVisitSummary, produces the aforementioned error. We verified that the necessary input_exposures (in this case, a calexp) exist on-disk and are butler-gettable.
Jim Bosch believes this is a pipeline or step-definition issue and suggests that we should allow isolatedStarAssociation in step2b to figure out all the required tracts by running it with a visit constraint (either explicit or from the input visits) and no tract constraint, and also passing skymap= explicitly. Since we are running step2b (which contains isolatedStarAssociation) with this data query: skymap='hsc_rings_v1' AND tract=9813, we can either add visit= and remove tract= and let it determine all necessary tracts OR move isolatedStarAssociation from step2b to step2a. We have chosen the latter as a solution for this ticket.
Attachments
Issue Links
- relates to
-
- Done
Activity
I relocated the isolatedStarAssociation task from step2b to step2a and initiated a sequence of bps runs via DM-36355. The runs were successful, indicating that isolatedStarAssociation is functioning as intended in step2a without any errors or failures. Subsequently, I executed step2b and step2d, which generated the anticipated outputs, successfully concluding all the required tasks without encountering any issues.
Here are the tasks that were executed from step 1 to step 2d in the new setup:
step1 tasks
|
------------------------------
|
characterizeImage
|
calibrate
|
isr
|
writePreSourceTable
|
transformPreSourceTable
|
|
|
step2a tasks
|
------------------------------
|
consolidateVisitSummary
|
consolidatePreSourceTable
|
isolatedStarAssociation
|
|
|
step2b tasks
|
------------------------------
|
jointcal
|
|
|
step2d tasks
|
------------------------------
|
finalizeCharacterization
|
writeRecalibratedSourceTable
|
updateVisitSummary
|
transformSourceTable
|
consolidateSourceTable
|
After updating the step definition, the pipeline visualizations for different steps are as follows.
Step 1
pipetask build \
|
-p $DRP_PIPE_DIR/pipelines/DECam/DRP-Merian.yaml#step1 \
|
--pipeline-dot /tmp/pipeline.dot; \
|
dot /tmp/pipeline.dot -Tpdf > /home/en7908/logs/DM-36355/pipeline_merian_step1.pdf
|
Step 2a
pipetask build \
|
-p $DRP_PIPE_DIR/pipelines/DECam/DRP-Merian.yaml#step2a \
|
--pipeline-dot /tmp/pipeline.dot; \
|
dot /tmp/pipeline.dot -Tpdf > /home/en7908/logs/DM-36355/pipeline_merian_step2a.pdf
|
Step 2b
pipetask build \
|
-p $DRP_PIPE_DIR/pipelines/DECam/DRP-Merian.yaml#step2b \
|
--pipeline-dot /tmp/pipeline.dot; \
|
dot /tmp/pipeline.dot -Tpdf > /home/en7908/logs/DM-36355/pipeline_merian_step2b.pdf
|
Step 2d
pipetask build \
|
-p $DRP_PIPE_DIR/pipelines/DECam/DRP-Merian.yaml#step2d \
|
--pipeline-dot /tmp/pipeline.dot; \
|
dot /tmp/pipeline.dot -Tpdf > /home/en7908/logs/DM-36355/pipeline_merian_step2d.pdf
|
Lee Kelvin, I wanted to ask if you could take a moment to review this ticket when you can. Your feedback would be much appreciated. Thanks!
Thanks Erfan, this looks good to me. Although I don't think Jenkins would be sensitive to the changes made here, I'd still recommend running this ticket through Jenkins to make sure nothing unexpected crops up. Otherwise, I think this looks good to merge to me - thanks!
We are going to edit https://github.com/lsst/drp_pipe/blob/main/ingredients/DECam/DRP.yaml