I noticed that there had been an update pushed out yesterday which included code changes as described above for the plate solving method. I've updated just now and will retry when the weather clears up.
I ran into a similar issue to this one last night when trying to us the polar align with platesolve. Can you confirm if your code change is integrated into 3.6.5? If yes, then I may have a different issue causing me to get the "Solver Failed" message. My experience was this:
1. Start Polar alignment
2. Capture all three images and plate solve successfully.
3. I would start the refresh and it would refresh twice, I would try to move the AZ knob and then get the dreaded "Solver Failed after..." message. What's worse is that there's no recovering from it without stopping and restarting the Polar Alignment process.
If your changes are already in my version, I will start a new thread for this.
That bugfix made it into 3.5.5 or 3.5.6, so way before what you are using.
Some failed solves in the refresh cycle shouldn't be an issue as long as it can solve some.
It isn't unusual to fail a solve right after turning a knob, as that rocks your mount/scope and the imaging isn't coordinated with that.
Hopefully it solves on the next one.
If you successfully solve the 3 exposures, and then have issues in refresh, it is often just the shorter exposures--try using a longer refresh period, e.g. 2s or 3s, can get you better images and a better chance to solve.
Not sure why you're saying that you have to start over. What makes you start over?
Are you using the "Plate Solve" technique (as opposed to the Move Star one)? If not, try that one.
Feel free to send a full debug log (certainly with "align module checked) if you're still having issues.
I am using the "Plate Solve" Method for the polar alignment. I would have to completely restart because after I see the first "Solver Failed..." message, it does not appear to do anything else. I think I need to do some more testing so that I can give you a better report of the issue. I'm hoping to have some clear skies soon so that I can test it. It is entirely possible that this is user error.
Just double checking that you clicked on the refresh button to get the process started.
After the initial 3 exposures that calculate the polar alignment error and required correction, you need to set a few parameters like the algorithm, the refresh interval, and then you click on refresh and the refresh cycle starts up.
OK. Well, I plate solved last night, and for me, though I'd often get a solver failed while adjusting the alt or az knobs, it would resume OK on the next exposure.
I used a 5s refresh exposure for my 250mm wide f/8 reflector.
Refresh times out after about 10s and takes another exposure. The iteration counter should increase on every exposure.
Send a debug log if you continue to have issues--make sure you have verbose logging and at least align module checked (if not all the modules--indi debugging isn't necessary).