The new-style kernels need to be copied from /project/czw/DM-30738/newBfKernels to /datasets/DC2/repoRun2.2i/CALIB/bfk. This will match the new mapper definition for bfKernel, which will now look with a template: bfk/bfKernel-%(raftName)s_%(detectorName)s.fits.
The ip_isr change was necessary to get the updates done on DM-30730 in runQuantum propagated to readIsrData.
Finally, this was tested with a template of ../rerun/private/czw/bfk/bfKernel-%(raftName)s_%(detectorName)s.fits to avoid various mapper issues. This successfully found and used the kernel
processCcd.isr INFO: Applying brighter-fatter correction using kernel type <class 'numpy.ndarray'> / gains <class 'NoneType'>.
|
processCcd.isr INFO: Finished brighter-fatter correction in 2 iterations.
|
but then crashed with a segfault at
processCcd.charImage.measurePsf.psfDeterminer WARN: NOT scaling kernelSize by stellar quadrupole moment, but using absolute value
|
processCcd.charImage.measurePsf INFO: PSF determination using 63/63 stars.
|
Caught signal 11, backtrace follows:
|
/home/czw/dev/DM-30738/utils/lib/libutils.so(+0x26389) [0x7f51f0eae389]
|
/lib64/libc.so.6(+0x36450) [0x7f51fb61d450]
|
/home/czw/dev/DM-30738/afw/python/lsst/afw/detection/_detection.so(+0x3b2c4) [0x7f51eaeaa2c4]
|
As I needed to rebuild five extra packages to get obs_lsst to build, I'm hoping this is just something in my environment being broken. An earlier segfault (in CR detection) was fixed with a rebuild of meas_algorithms, afw, and utils from the current master.
Jenkins build to check that HSC BF correction is unaffected here: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/34458/pipeline
Improvement
DM-30730 DC2 Reprocessing with w_2021_24 (gen2)
The new-style kernels need to be copied from /project/czw/
DM-30738/newBfKernels to /datasets/DC2/repoRun2.2i/CALIB/bfk. This will match the new mapper definition for bfKernel, which will now look with a template: bfk/bfKernel-%(raftName)s_%(detectorName)s.fits.The ip_isr change was necessary to get the updates done on
DM-30730in runQuantum propagated to readIsrData.Finally, this was tested with a template of ../rerun/private/czw/bfk/bfKernel-%(raftName)s_%(detectorName)s.fits to avoid various mapper issues. This successfully found and used the kernel
but then crashed with a segfault at
processCcd.charImage.measurePsf.psfDeterminer WARN: NOT scaling kernelSize by stellar quadrupole moment, but using absolute valueAs I needed to rebuild five extra packages to get obs_lsst to build, I'm hoping this is just something in my environment being broken. An earlier segfault (in CR detection) was fixed with a rebuild of meas_algorithms, afw, and utils from the current master.
Jenkins build to check that HSC BF correction is unaffected here: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/34458/pipeline