Progressive OOM issue with SGP

Sequence Generator
1

Hello SGP Team,

For the last several weeks, I’ve been having some OutOfMemory issues with SGP. It seems to only affect the 2.6.x versions, as I managed to get a hold of some 2.5.x versions from other SGP users, and the issues did not present.

I picked up SGP 2.6.0.13 yesterday, and did some testing, and at first it appeared as though the issue was not presenting. However, sadly I just started receiving the errors again.

This is the error, consistent across all o the 2.6.x versions I’ve tested (which was everything on the download page), including the 0.13 release:

[02/20/17 16:14:43.014][DEBUG] [Camera Thread] SetAscomNormalSpeed...
[02/20/17 16:14:43.014][DEBUG] [Camera Thread] Cannot set readout speed, not supported by camera...
[02/20/17 16:14:44.030][DEBUG] [Camera Thread] ASCOM Camera Error : CheckDotNetExceptions ASCOM.ASICamera2.Camera ImageArray Get System.OutOfMemoryException: Insufficient memory to continue the execution of the program. (See Inner Exception for details) (System.OutOfMemoryException: Insufficient memory to continue the execution of the program.)
   at ASCOM.DriverAccess.MemberFactory.CheckDotNetExceptions(String memberName, Exception e) in c:\ASCOM Build\Export\ASCOM.DriverAccess\MemberFactory.cs:line 616
   at ASCOM.DriverAccess.MemberFactory.GetTargetInvocationExceptionHandler(String memberName, Exception e) in c:\ASCOM Build\Export\ASCOM.DriverAccess\MemberFactory.cs:line 650
   at ASCOM.DriverAccess.MemberFactory.CallMember(Int32 memberCode, String memberName, Type[] parameterTypes, Object[] parms) in c:\ASCOM Build\Export\ASCOM.DriverAccess\MemberFactory.cs:line 288
   at ASCOM.DriverAccess.Camera.get_ImageArray() in c:\ASCOM Build\Export\ASCOM.DriverAccess\Camera.cs:line 361
   at qo.h7(oz A_0, qy& A_1)
[02/20/17 16:14:44.030][DEBUG] [Camera Thread] ASCOM camera: frame and focus error. : Error capturing during frame and focus!
   at qo.gp(oz A_0, qy& A_1)
[02/20/17 16:14:46.386][DEBUG] [Camera Thread] ASCOM camera: frame and focus...

When the error presents, it begins a downward spiral. The error will present again and again, and more and more frequently, until it occurs on every frame downloaded (regardless of how it is downloaded.) Interestingly, it seems to always be F&F/AF frames that initially trigger the issue…I cannot recall any instance where a normal sequence frame download ever initiated the issue.

I am using an ASI1600MM-Cool. I originally posted the error to the ZWO guys. They came back stating that this issue occurred once before with SGP, and that it had been corrected (and it must have occurred before 2.5.2.5, as that was the earliest version of 2.5.x that I’ve tested with).

I’ve done extensive testing of the camera and several ASI ASCOM driver versions with several other programs, including SharpCap, FireCapture and APT. I have had no issues with any of those programs, even after extended hours of use. I let SharpCap pull video from the camera for about two days strait without interruption, with periodic runs of PHD2, as well as periodic fiddling with the focuser and filter wheel. I experienced no issues, with either the ASCOM driver or DirectShow driver, under any circumstances, not even a dropped frame.

The issue seems to be pretty specific to SGP 2.6.x. It did seem to get better with 2.6.0.13, as I was able to acquire F&F frames continuously for about an hour before I finally encountered the OOM error, which was much longer than the 5-15 minutes it took with beta versions.

2

Jon,
This was reported in another thread. Please see here.

Thanks,
Jared

3

Thanks for the reply, Jared. I did read that other thread.

Just to make sure we are on the same page, I have encountered this error with 2.6.0.13, the final release of SGP 2.6 that was made available yesterday. Is the LBA issue still present in the final release version?

4

It is. We’re working to correct it now. We’ll likely put out another release to address it in the next 48 hours.

Thanks
Jared

5

Gocha. Thanks for the info, Jared! Very much looking forward to the next release.

6

Just to back Jon up again, I had this happen with 2.6.0.13 last night as well. Nowhere as frequently as 2.6.0.12 though…that build certainly had some issues. :slight_smile:

2.6.0.13 only gave me 3 instances of this issue over an imaging run that spanned 6 hours including a very nicely executed meridian flip.

I did not have a chance to test with 2.6.0 and the forecast is now for clouds for a few days.

Looking forward to the 2.6.0.x update guys. Very much appreciated.