I've now had a few issues when running a job from the scheduler that has led to a telescope crash. 

The setup:

• I have a capture sequence in scheduler to capture HaLRGB, which is fairly short and I typically cycle through it multiple times.
• The schedule is set up to run this sequence repeatedly until 4am (using the run until checkbox)
• Astroberry on RbPi 4, 8gb data to an external SSD
• PhD2 Guiding
• Meridian flip enabled
• Watchdog enabled
• Parking is setup only through the scheduler

The Bug:

• At some point when restarting the sequence Ekos will report "Cannot Open the Sequence ha_lrgb_sequence.esq"
• It will then spam this warning endlessly, I believe every time it tries to restart the capture sequence. 
• The mount/guiding continues, but the meridian flip does not occur.
• The completed sequence procedure (park the mount) does not occur.
• The telescope crashes.  
• The log stops "happily" [2021-04-19T00:52:05.592 PDT INFO ][   org.kde.kstars.ekos.capture] - "Received image 19 out of 20."
• It is difficult to debug because I have to click through what seems to be an infinite number of the popups (I haven't been able to get through all of them)

This has happened two nights in a row.

I don't know that I consider the file read a bug in and of itself.  That could be the sd card on the RbPi.  The bug as I see it is 

1. The system loops forever and never terminates
2. In the process of doing this the meridian flip and the shutdown and the sequence are ignored
3. I would much rather the system go failsafe when this error occurs than to continue forever and potentially ruin my equipment!

I'm happy to assist in any way to help debug etc, although my skills here are limited.

Hi Tony! Welcome to INDI forums. Thank you for your detailed report.

Regarding the Watchdog, this would not be very useful in you're running the server locally on raspberry pi. This is meant for remote connection. i.e. if INDI server loses communication with a remote client, then it triggers the watchdog.

For the sequence error, I submitted an MR that should help with this rare situation: invent.kde.org/education/kstars/-/merge_requests/273

However, there is still some discussions on perhaps a global mode where popups are completely suppressed.

Is there a preferred way to implement a HA "safety" I have been tempted to set up an independent INDI client to check the mount position and force a stop if it goes out of range and KSTARS has ceased responding.