D19528: Making Scheduler robust against guiding problems

[Wolfgang Reissenberger]

wreissenberger created this revision.
wreissenberger added reviewers: mutlaqja, TallFurryMan.
Herald added a project: KDE Edu.
Herald added a subscriber: kde-edu.
wreissenberger requested review of this revision.

REVISION SUMMARY
  In the current implementation it could happen, that guiding problems bring the Scheduler and the Capture module out of sync. If capturing is suspended, a subsequent guiding error remains undetected. As a consequence, capturing remains suspended and guiding will not be restarted.
  
  This situation is not untypical during a night with high cirrus that occasionally disturb guiding. First they lead to guiding deviations and from time to time to lost guiding stars. In such nights it could happen, that a single cirrus cloud interrupts the entire capturing.
  
  In order to cover this, the behaviour of Capture and Scheduler is changed:
  
  - Aborted guiding leads to aborted capturing even when capturing is already suspended.
  - Repeated guiding problems lead to jobs being aborted, not to be marked with errors.
  - In order to avoid aborted jobs to be immediately restarted, the scheduler considers aborted jobs for restart, if all potentially executable jobs have been aborted.
  
  Additionally, two other weaknesses have been resolved:
  
  - Removed restarting aborted capture when guiding resumes, since it conflicts with the Scheduler.
  - In the PHD2 adapter, connectEquipment() may be called arbirtrarily often when equipment is connected.

TEST PLAN
  - Create a schedule with at least two schedules, that may both be executed and start the scheduler with guiding enabled. The schedules should use an imaging plan with a narrow guiding deviation set (i.e. 1 arcsec)
  - Change the guiding parameters such that guiding is bad and creates a guiding deviation above the limit.
  - Interrupt the guiding.
  - The scheduler should restart now guiding and then capturing.
  - Interrupt the guiding for another 4 times.
  - After the last interruption, the scheduler should set the schedule to ABORT ad take the next schedule.
  - After 5 guiding interruptions, the second schedule should be set to ABORT.
  - When all schedules are ABORTed, one they should all be set to SCHEDULED and one of them being restarted.

REPOSITORY
  R321 KStars

BRANCH
  EKOS/robust_guiding

REVISION DETAIL
  https://phabricator.kde.org/D19528

AFFECTED FILES
  kstars/ekos/capture/capture.cpp
  kstars/ekos/capture/capture.h
  kstars/ekos/guide/externalguide/phd2.cpp
  kstars/ekos/scheduler/scheduler.cpp

To: wreissenberger, mutlaqja, TallFurryMan
Cc: kde-edu, narvaez, apol
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-edu/attachments/20190304/894cd162/attachment-0001.html>