The online solver does what you described quite often. It sometimes works and sometimes doesn't. I would say at least 3/4 of the time it does not work. Either it fails with the job id issue or the server is down entirely.

Now the offline solver on the other hand, works very very well once you get it configured properly. I spent more than a week trying to get the settings just right. When it is configured correctly, it should solve in less than 10 seconds, if it is not then it might take many minutes and still not solve.

I found a couple of tricks to be helpful, you might find them useful

1. You mentioned a problem with noise and said you are already binning. I had a problem like that in my Canon camera. i have found that you can get rid of some noise and decrease the number of stars found by using the downsample option like this: --downsample 4. Also, if there is a lot of noise, taking a slightly longer photo sometimes finds fewer stars. (This is counterintuitive because a longer exposure should actually find more stars, but if the "stars" are actually noise then a longer exposure seems to help) You can tell if you have it correct if you turn on the verbose option in the solver in Ekos and look at how many stars it finds. If it finds several thousand, you should increase binning/downsampling and you should tweak the exposure time either longer or shorter. These options will also make it solve much quicker because you can get it to look for fewer stars. I found about a hundred or so to be a good target to shoot for.

2. I found that sometimes the options about the frame size (for example: -L 19.3223 -H 26.6952 -u aw) actually can prevent it from solving. Sometimes I find it is better with those options and sometimes without. If your images are not solving, try taking that part out.

3. One of the most important things is getting the right index files. The website says 10% to 100% of your frame size, but I have found this to be excessive. What I did is I took several test images in different parts of the sky with my scope, saved them to my computer and uploaded them to the website nova.astrometry.net. Then I paid attention to which index file actually solved each image. One thing I found that was surprising was that sometimes it solved with files in the 4200 series and sometimes it solved with files in the 4100 series. Then I downloaded files in the same range as those that solved my test images. For my particular scope (a 6 inch f4, an SBIG and a paracorr), I found that the 4207 - 4210 files and the 4107 – 4110 files were what I needed. I found that images using the simulator usually solved with the 4205 index files. Obviously if you switch to a different scope or a different camera then different files or settings will work best

Hopefully this helps,

Rob

Hey Krno. Thanks for the hard work !!!

I tested it last night. It works very well. He recognized every time a real star instead of a hot pixel. The same thing with the focusing.

In my experience the solver can fail for a few reasons:
1) error during the exposure, giving an image with guiding errors. Then he will try again and again to solve on the bad image. A solution for this would be to slew a few arcminutes away and then try a 2nd or even 3th time. Robotic observing means that you should be able to catch as many error as possible while you're not looking on the scope.
2) offline solver does not find a solution. A solution would be to do an autostretch, convert to JPEG and send it to the online solver.
3) The previous goto was completely off. Then the offline solver will probably timeout. Sending the image 2nd stage to the online solver would also catch this error.

I haven't experimented much with the index files yet. However, I installed all of them. I found that with radius=30degrees and the standard settings, the offline solver found every time the solution pretty fast. However, I come from CCD-commander and there it can work a lot, until it fails a few times and you are completely in the dark about the source of error. The atmospheric conditions are important, guiding errors, wind...