hpc4fun
commented
2 years ago More info: In the builds above I did clean the build folder before each rebuild and for the latter case the process was repeated a couple of time (clean build folder, rebuild, install).
I took the time to report the issue above and while doing that I wiped the clone and did a fresh clone (I did the same for the first phone above but not for the second one where I assumed that a clean build folder was sufficient). It seems that it wasn't sufficient as a rebuild from a fresh clone apparently made the NaNs disappear and the Loop crashes has now stopped (as least for now :)).
One may speculate if the NaNs simply disappeared due to the time that has went by since I started to debug this (as the arrays are finite one could imagine that they would be filled with sane values eventually) or was the true fix to re-clone and start from scratch. That I don't know.
The Loop app has been rock solid for years (several versions) so the main issue here is why do the apps suddenly crash out of the blue and maybe not so much the hassle to get the communication with the pod reestablished. It shall be stressed that this is not the dev version but the current master so the sudden crashes come as a true surprise to me. It should also be mentioned that I have issue reports from before (previous pods) and after the crashes on both the cases described above so just reach out if more details would be of interest.
marionbarker
ps2
The other day I had a loop app that crashed when one tried to open it. The symptoms looked very much like the one described here: https://loopkit.github.io/loopdocs/troubleshooting/loop-crashing/ but from the loop issue report we see:
so the issue is not due to an expired profiled. I managed to regain access to the loop app and the running pod by rebuilding the loop app:
The running pod was not lost and I continued running with it. The pod has been changed since but the loop app has not crashed again.
Now, today (or rather during the night :) another T1DM running loop had the same kind of crash. The loop app was crashing when one tried to open it and there was no way to talk to the app nor the pod. I thought the fix was easy. Simply redo the process above but alas it turned out that there was more to it this time.
I did rebuild and reinstall again. This time from:
to
Again this trick allowed me to regain control of the app (I could now start it and get an issue report out). Unfortunately this time, the loop app would crashes on a regular basis and the predicted BG is NaN. Regular in the sense that it was easy to make it crash. I trust that the crash is due to the fact that there are NaNs in internal data structures (for some reason I don't understand) and when these are being used in computations it will crash, e.g. I saw one of the RL subs that was trying to convert this double to int and consequently crashed in trying to do this conversion.
In the issue report I see the NaNs too:
appGroupName: group.com.FT54B2L35J.loopkit.LoopGroup statusExtensionContext: Optional(["predictedGlucose": ["interval": 300.0, "unit": "mmol<180.1558800000541>/L", "values": [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan], "startDate": 2022-01-22 01:50:00 +0000], "reservoirCapacity": 200.0, "netBasal": ["percentage": 0.0, "start": 2022-01-22 01:05:42 +0000, "rate": 0.0], "lastLoopCompleted": 2022-01-22 01:18:27 +0000, "pumpManagerHUDViewsContext": ["pumpManagerHUDViewsRawValue": ["hudProviderViews": { alerts = ( ); lifetime = "259146.8027609587"; podActivatedAt = "2022-01-21 16:56:01 +0000"; validTime = "2022-01-22 01:49:31 +0000"; }, "managerIdentifier": Omnipod]], "version": 5])
If I try to enter any of the menus (blood glucose, active insulin, ...) it consistently crashes the loop app.
Fortunately, I can still bolus and the basal seems to run OK so I can get through the night in open loop mode but a proper fix is obviously needed in order to continue.
Phone in the first case is latest SE model whereas it's the older SE model in the second case. Both phones run ios 15.2 and there is sufficient storage on both phones.
ANY feedback or ideas on how to proceed from here is highly appreciated.
Loop Report 2022-01-22 02.49.41+01.00.md
PS! I should also mention that in the first case (the best case) the pod was on its third day whereas in the second case the pod was activated the day before, i.e. had less than 24 hours of operation when the Loop app crashed.