| | 346 | * TdB: In an effort to automate the desperate diff generation, a revised nightly_science.pl script has been made. It is called nightly_science_desdifftest.pl and found under ~ippitc for the time being. The main change is that a new subroutine is created, called desperate_diff_singles which is called after the regular queue_diffs subroutine when the configuration of a particular label has desperate diffs turned on. This subroutine is much like the desperate_diff_queue subroutine that gets executed at the end of a night, but instead queues up desperate diffs for a given object on the fly when the following conditions are met: |
| | 347 | {{{ |
| | 348 | -it has an uneven number of warp exposures |
| | 349 | -the time since the last exposure is at least x hours, where x is set as an input parameter called desdiffdt in units of hours (The default is 3 hours) |
| | 350 | }}} |
| | 351 | An example call to the script is: |
| | 352 | {{{ |
| | 353 | nightly_science_desdifftest.pl --dbname gpc1 --camera GPC1 --verbose --check_diffs --date 2019-05-28 --desdiffdt 3. --debug |
| | 354 | }}} |
| | 355 | For the time being, we should test the workings of the script. A crontab has been added on ippc30 which runs the script every 3 hours and pipes the output of the check_diffs for both gpc1 and gpc2 to the ps-ipp-ops mailing list. Furthermore, for the czars, they should try out desperate diff queuing using the new way by picking an appropriate desdiffdt time when the first chunk is done and incomplete. A word of caution: '''Do not use a desdiffdt that is too short!''' The timescale is there by virtue of not having a more sensible way of judging whether a chunk is complete and will not be resumed/picked up. Picking a very low desdiffdt could lead to unforeseen circumstanced like queuing desperate diffs for chunks still in progress or temporarily interrupted. |
| | 356 | |
