×

INDI Library v2.0.7 is Released (01 Apr 2024)

Bi-monthly release with minor bug fixes and improvements

Scheduler additions: Group Repeat & Repeat All Jobs

  • Posts: 1224
  • Thank you received: 566
This note is to let you know about a couple of additions to the scheduler: Group Repeats, and Repeat All Jobs. They were recently added to the latest software, so should be available in the nightly releases soon. They will be a part of the next release (3.6.6). As always, feedback is welcome.

Group Repeats (only available with the Greedy scheduling algorithm)

This new feature allows you to run two or more Scheduler jobs at roughly the same priority, such that if they were both runnable, they would progress at roughly the same rate.  This may be applicable, for example, to jobs imaging multiple tiles in a mosaic, but is generally applicable to any set of jobs.

Normally the (Greedy) scheduler's job priority is set by the row a job is listed on in the Scheduler's job table. Jobs on rows closer to the top run with higher priority than jobs lower down. Thus, if a job on row 2 (Job2) uses Repeat Until Terminated, and that job's running constraints are satisfied, a lower down job (e.g. Job3) will not be scheduled to run. 
If you wish to alternate jobs, you can now assign each of the jobs the same group name, and give the jobs one of the repeating finish conditions (e.g. Repeat for N times, or Repeat Until Terminated). With that setup, jobs in the same group will cede to each other if they have completed more 'Repeat Iterations' than the other job. So, if Job2 with group "MyGroup" completes its 2nd iteration, and Job3 with the same group name has only completed 0 or 1 iterations, when the time comes to schedule Job2, Job3 will run instead.

Practically speaking, imagine you had a 6-panel mosaic you wanted to alternate. You might give all of those jobs the same group name, and make them all e.g. "Repeat for 5 times". Then, they would run in lock-step. The cadence of job switching would be controlled by the length of the sequence file assigned to each of those jobs. You wouldn't want to make the cadence too short (e.g. capturing one 2-minute image), as there is overhead in switching jobs. For instance, starting jobs may involved aligning, starting guiding, and even focusing.

Adding group names to all jobs in a mosaic is now an option in the Mosaic Planner.          

The arrow below points to the new group name input box.




Repeat All Jobs (only available without Remember Job Progress)

Wolfgang recently added  a checkbox and number input right below the Scheduler jobs table that allows you to repeat the entire schedule N times.  This can be used to alternate a few jobs. You can just list the jobs on the scheduler, set it to repeat N times, and the jobs will repeat. However, this change is incompatible with 'Remember job progress (below) and unavailable if Remember job progress is checked. (Note: Remember job progress is recommended.)





Remember job progress

This is not a new feature, but I wanted to remind folks why it is there, and that it's recommended for typical use--if you are saving your image files to the computer where KStars/Ekos is running (which is often the case). When this box is checked and a job is running, the Scheduler looks at the job's images already captured on disk, and doesn't re-capture the ones that are already there.  The benefit is that if a job is restarted, or if multiple jobs are being run together, stopping and starting when they are runnable, then jobs re-start right where they left off. If Remember job progress is unchecked, jobs would restart from the start of the sequence specification each time, which probably isn't what you want.  Unfortunately, if the images are not stored on the same computer that Ekos is running on, then this feature doesn't work and the jobs restart from their beginning.

A possibly confusing side-effect of Remember job progress is that if you've run a job using the (default) Sequence Completion finish condition, and it has captured all its images, and now you want to run the job again, the Scheduler won't schedule the job because it believes that all the images have already been captured. You'd either need to move those images elsewhere on disk, or change the finish condition to Repeat for N Times, or Repeat Until Terminated.

The following user(s) said Thank You: Jasem Mutlaq, Rafael Schlegel, AstroMuni, John, Juergen_from_Bavaria
9 months 3 weeks ago #93797
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 969
  • Thank you received: 94
Thanks Hy. Nice updates.

"You'd either need to move those images elsewhere on disk, or change the finish condition to Repeat for N Times, or Repeat Until Terminated."

... or, run the job under a different name as many times as you need:
m39-1
m39-2
...
m39-n

Cheers and clear skies,
Steve
Last edit: 9 months 3 weeks ago by alacant.
9 months 3 weeks ago #93832

Please Log in or Create an account to join the conversation.

  • Posts: 41
  • Thank you received: 1
Hi, i have find a little bug when rise the meridian flip rutine:

Tonight i was capturing a 2x4 mosaic repeating all jobs 5 times, over NGC7000.

At 04:46 hour, start the meridian flip,
2023-07-07T04:37:11 Meridian flip completed.
2023-07-07T04:36:18 Meridian flip running...
2023-07-07T04:36:17 Meridian flip ready to start...

just when finish the last part from mosaic in 3rd iteration like said this log:
2023-07-07T04:36:18 No jobs left in the scheduler queue.
2023-07-07T04:36:17 Job 'NGC7000-Part_8' is complete.
2023-07-07T04:26:09 Job 'NGC7000-Part_8' capture is in progress...

During meridian flips working, the scheduler start the 4 iteration, and display a error from 4 parts:
2023-07-07T04:36:30 Job 'NGC7000-Part_4' is terminated due to errors.
2023-07-07T04:36:30 Warning: job 'NGC7000-Part_4' slew failed, marking terminated due to errors.
2023-07-07T04:36:29 Job 'NGC7000-Part_4' is slewing to target.
2023-07-07T04:36:27 Job 'NGC7000-Part_3' is terminated due to errors.
2023-07-07T04:36:27 Warning: job 'NGC7000-Part_3' slew failed, marking terminated due to errors.
2023-07-07T04:36:26 Job 'NGC7000-Part_3' is slewing to target.
2023-07-07T04:36:24 Job 'NGC7000-Part_2' is terminated due to errors.
2023-07-07T04:36:24 Warning: job 'NGC7000-Part_2' slew failed, marking terminated due to errors.
2023-07-07T04:36:23 Job 'NGC7000-Part_2' is slewing to target.
2023-07-07T04:36:21 Job 'NGC7000-Part_1' is terminated due to errors.
2023-07-07T04:36:21 Warning: job 'NGC7000-Part_1' slew failed, marking terminated due to errors.
2023-07-07T04:36:20 Job 'NGC7000-Part_1' is slewing to target.
2023-07-07T04:36:18 Starting job sequence iteration #4

when meridian flip complete, slew the part5 and start capture procedure:
2023-07-07T04:39:25 Job 'NGC7000-Part_5' capture is in progress...
2023-07-07T04:39:25 Job 'NGC7000-Part_5' guiding is in progress.
2023-07-07T04:37:48 Starting guiding procedure for NGC7000-Part_5 ...
2023-07-07T04:37:47 Job 'NGC7000-Part_5' repositioning is complete.
2023-07-07T04:37:47 Job 'NGC7000-Part_5' alignment is complete.
2023-07-07T04:37:11 Job 'NGC7000-Part_5' is capturing and plate solving.
2023-07-07T04:37:11 Job 'NGC7000-Part_5' slew is complete.
2023-07-07T04:36:32 Job 'NGC7000-Part_5' is slewing to target.

@Wolfgang,
@Hy Murveit
Please, can you check this? maybe can be a possible bug.

Thanks,
Regards.
9 months 2 weeks ago #93857

Please Log in or Create an account to join the conversation.

  • Posts: 41
  • Thank you received: 1
Sorry, the Bug Captured
Last edit: 9 months 2 weeks ago by Javier Albella.
9 months 2 weeks ago #93858
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 1224
  • Thank you received: 566
Javier,

Could you please post the full log somewhere, or repeat and get the log.
Please make sure the logs are debug logs--that is, set up like the screenshot below with Verbose on, set to log to file, and all ekos modules checked.
If it is too large to post, please upload to Google Drive or DropBox and send the link.

Also, please also post the .esq and .esl files (and assuming we should still use Grand Canaria).

If I were you, I would set these up as "repeat for 5 iterations" jobs, not check the "repeat all jobs" checkbox,
and assuming you are saving the files on the same machine as Ekos is running, then check "Remember Job Progress".
Then any failed jobs can catch up later. Perhaps you also could consider setting the "Abort Job Management" for each tile
job to "reschedule errors 5 minutes" or some such.

All that said, personally, I would be surprised if the "slew failed" errors are related to the scheduler changes.
I'm not as up-to-speed on mount issues as e.g. Wolfgang, but it might be interesting to see what happens if you set up some job that
slews among all the mosaic tiles right at the meridian. Perhaps the issue is related to that?

Hy
9 months 2 weeks ago #93859

Please Log in or Create an account to join the conversation.

  • Posts: 41
  • Thank you received: 1
Hello Hy Murveit,

i tried to test with simulator, but i do not reproduce the bug again, here are the log , esq and esl files and some captures from this test. In this case I made a mosaic 2x4 over M31 with 600seg by capture.

Tonight i try again but in a real scenery.

Thanks,
Regards.
9 months 2 weeks ago #93868

Please Log in or Create an account to join the conversation.

  • Posts: 41
  • Thank you received: 1
Hi again,
Yesterday i did a real scenery to repeat the bug, but the meridian flip and all subrutine went ok.
I can not reproduce again, so its an isolated case.

I upload the test with log file here:

File Attachment:

File Name: NGC7000_Mosaic2x4.zip
File Size:4,119 KB
Last edit: 9 months 2 weeks ago by Javier Albella.
9 months 2 weeks ago #93906
Attachments:

Please Log in or Create an account to join the conversation.

Time to create page: 0.937 seconds