Sean Pickard created a new topic ' Memory Leak' in the forum. 4 years ago

I had multiple crashes over the night due to running out of memory. I was watching the system monitor as kstars crashed a few times. Below I have listed all the components that I am using. I looked up similar issues and work arounds to them. Have already disabled fits view as that was causing me issues previously. After I had disabled that I was able to image for 3 nights in a row (last time it was clear) with no crashes all night. This time when I booted up it would crash after taking 4-10 captures. In watching the system monitor I started my process and it would go through slew, focus, align, guide, and start first capture all fine. Memory would spike during each operation but then return to its original amount. When it got to finishing a capture, downloading, and beginning the next capture however the memory would increase by ~20 MiB each time until it ran out of memory when downloading a capture and kstars would force close. I ran kstars in gdb using

gdb -ex run kstars
as recommended by Jasem in this thread.
(gdb) bt
returned "No Entry" I have included the syslogs, and a screenshot from Stellarmate App from when I disconnected the drivers and closed KStars, and another from when I reopened KStars. I know a "temporary fix" would be to upgrade to the Raspberry Pi 4 - 4gig. But if the memory leak is still there, it will still be an issue, just would take longer between crashes.

File Attachment:

File Name: 2019-11-07.zip
File Size: 948 KB


Using:
Rasberry Pi 3b+ w/ Stellarmate OS v.1.4.4
EQ6-R pro with EQMOD
Canon EOS Rebel T6 imaging camera
ZWO ASI290mm mini guide camera
USB GPS
ZWO EAF

Read More...

Sean Pickard replied to the topic 'Issues' in the forum. 4 years ago

Sorry, I had that in the list of things I needed to put in my post, but forgot.

Raspberry Pi 3b+
Whenever I looked it was around 50%
I have DSLR image preview popup in minimal mode, and disabled as I had seen that was a solution to crashing awhile ago. Fixed the crashing under stable build.

Read More...

Sean Pickard created a new topic ' Issues' in the forum. 4 years ago

Ran into a few issues last night. Updated to nightly build to see if it would fix the sequencer start and meridian flip issue that I've had in the past. Started the sequencer 30m before twilight to see if the sequence would start correctly. It did not. I waited for about 15m after to see if it would start due to internal time being off. I am using a USB GPS to update KStars/Ekos with time and location. Had issues with focuser as I had left it out at previous location when I shut down for the morning a while ago and step position was reset. Taking my laptop outside with ASICAP, I reset that to initial. I was able to plate solve to M33 by using the guide camera, but I was off by 15 arcmin or so when attempting to plate solve to M33 with my imaging camera. I did not have this issue when I was imaging M31, but have had this issue on multiple nights trying to capture M33. Synced the plate solve of the guide scope with imaging scope by repeatedly plate solving and adjusting the guide scope until they were within close to being centered. Then went to image M33 using the guide scope, was near the meridian so had issues getting aligned. Ended up using mount control function to get M33 close to center on the guide camera, then started an imaging sequence with use focus and guide enabled. After 4 captures, KStars crashed, used one of the images to plate solve and align with and used the guide scope to plate solve. Correctly aligned and plate solved. Another 6 captures and crashed again, was asleep by this point. Only seemed to crash during image capture sequences. The proceeding roughly 3 hours of tweaking and getting everything aligned I didn't have a crash. Attached is the sequence file, capture sequence file, full log files.

Using:
EQ6-R pro with EQMOD
Canon EOS Rebel T6 imaging camera
ZWO ASI290mm mini guide camera
USB GPS
ZWO EAF

Read More...

Sean Pickard replied to the topic 'A few issues' in the forum. 4 years ago

Oh, yes, that may happen as well when you are too close to the meridian and plate solving sends you across. But anyway, it looks like mount and KStars/EKOS do not have the exact same time or location. What would help in this case is adding 15min extra past meridian.[/quote]

I tend to have the meridian flip set to perform if HA is > 0.10

As for the known bug solution. I will try the "Repeat for 1 run" option next time the weather is clear. Which might be awhile, my electronic focuser is set to arrive tomorrow.

Wolfgang, do you have an idea when the nightly builds will have the fix?

Read More...

Sean Pickard replied to the topic 'A few issues' in the forum. 4 years ago

Megiddo wrote:

knro wrote: [2019-10-13T00:57:27.665 EDT INFO ][ org.kde.kstars.ekos.capture] - "No pending jobs found. Please add a job to the sequence queue."


I had that same error too. I was using the "Guiding Deviation" turned on and was using <2.00 (default). And that number was being hit and it canceled the job (putting it in Pending Status or something like that). It would go to the next job, do the same and put 5+ jobs into the Pending Status before kicking off again.

Once it got to the bottom of the jobs, it gave me the error: "No pending jobs found. Please add a job to the sequence queue."

I upped the Guiding Deviation to 3.00 and got more success, eventually going to 5.00.

I also started using Sequences of 2 shots vs 10+. It required many more entries of course, but if the failed jobs halted, it had plenty more to find.


I have had something like that before, I increased guiding deviation to 3.00 as well, made sure I had polar alignment spot on and enabled the re-schedule errors and would change job completion conditions to repeat until terminated.

Read More...

Sean Pickard replied to the topic 'A few issues' in the forum. 4 years ago

sterne-jaeger wrote: [post=44596]

A meridian flip is not a specific command to the mount but it simply re-issues a slew command to a target that has crossed the meridian since the last time it has been slewed to.

-Wolfgang


To expand on what happened as I don't have logs. I would be tracking the object for a few minutes, it would pass the meridian, pass the constraint I set, capture would finish. Mount would slew and plate solve. Sometimes it only slewed a few degrees, once it flipped, plate solved and flipped back.

Read More...

Sean Pickard replied to the topic 'A few issues' in the forum. 4 years ago

I didn't notice that I was able to save the scheduler before, so this is the first time that I saved the scheduler. Normally I would just create the session each day fresh. Attached is how the scheduler is normally configured with the capture sequence I used.

When doing testing during the day, I connected EQMOD and GPS and ran simulated guide camera and ccd camera. I would find an object about to pass the meridian, tell the scheduler to start imaging, sometimes it would say "performing meridian flip" only slew a couple degrees, plate solve and slew back. When an object that was passing the meridian near zenith, the mount flipped properly, plate solved and flipped back to the other side. I do not have logs of these happening, but could attempt to reproduce if necessary.

File Attachment:

File Name: Sequence_scheduler.zip
File Size: 1 KB


Read More...

Sean Pickard created a new topic ' A few issues' in the forum. 4 years ago

There are a few issues that I have been having with ekos that I have been working on.
Attached equipment is below:
EQ6-R pro with EQMOD
Canon EOS Rebel T6 imaging camera
ZWO ASI290mm mini guide camera
USB GPS (I am not near a time zone change for this to be an issue, KStars & Ekos are set to update from USB GPS)

The biggest issue is that the meridian flip function reports "waiting" but never actually performs the flip. I know that the function is set to wait for the current capture to complete, then perform the flip. In the attached logs, the flip is supposed to occur around 1am. I have included all modules that I am using as when I was doing some testing during the day with ccd and guiding simulator, the mount would perform a meridian flip. I have an EQ6R-pro with EQMOD. The way I have gotten around this issue is watching the failure happening, parking the mount and restarting the scheduler. The mount then slews to the correct side and continues tracking fine.

File Attachment:

File Name: log_23-35-45.zip
File Size: 5,374 KB


The next issue, which I do not currently have logs for is for the scheduler not starting the job process when I initialize the scheduler earlier on in the day and it is waiting for twilight. I adjusted the setting for pre-emptive shutdown to be longer than 2 hours but when the time comes for the job to start, nothing happens and the scheduler still reports "waiting for job start time".

When I open up the log directory from ekos to verify the destination, another copy of KStars is started and logging starts on that instance.

Read More...