What can slow my oref0 0.7.0?

21 views
Skip to first unread message

Petr Borodin

unread,
Jan 3, 2020, 4:05:36 PM1/3/20
to OpenAPS Dev
Hi all.
I'm using RPi with Pi HAT, MMT-722 pump, nightscout is hosted on Heroku. My CGM is G5 with Android phone.
I have a problem with the loop speed, all operations called 'refresh' (i.e. 'Temp refresh', 'meal.json refresh') take about 2 minutes each. So in the looplog ('l' in bash) i see Preflight (fast), the oref begins to update the data, it writes 'Temp refresh' and then freezes for couple of minutes. Same goes with meal.json refresh and so on.
I tired to do the next thing: SD card changed 3 times, it's totally new, class 10 Kingston and Sandisc. Next, I did a clen install of nightscout, cleaned all useless data. Next, i took another Pi board, new and shining, also i tried both overclocking to 1100 MHz and underclocking to 700 MHz. I should mean that I have a copper heatsink on the board. I did everuthing that I was able with my pump - enabled and disabled remote control and wireless meter (I've read somewhere it could help), did the self-testing 100 times. My WiFi at home is pretty quick, Heroku sites open less than in 1 sec. I use Android for CGM and tethering, it is also clean and not used for nothing more. My oref0 install is fresh and has no errors, so does the raspbian - it's up to date and clean.
Due to this problem of loop taking about 4-6 minutes I have to wait for the first bolus after carbs entering about 15-20 minutes. Also the prediction graph in NS looks very unclean.
What can I do with all this?
Thank you in advance.

Scott Leibrand

unread,
Jan 3, 2020, 4:08:38 PM1/3/20
to OpenAPS Dev
Make sure you're running the latest oref0: it's significantly faster than 0.6.x

You can run an OREF0_DEBUG=1 loop as described at the end of https://openaps.readthedocs.io/en/latest/docs/Troubleshooting/oref0-setup-troubleshooting.html to see exactly what your rig is doing in between normal print statements.

There is some active discussion at https://github.com/openaps/oref0/issues/1348 of improving the speed of oref0 on the Pi Zero.

You should also consider getting an Intel Edison (off eBay or similar) and and Explorer Board: it's much faster than the Pi Zero.

-Scott

--
You received this message because you are subscribed to the Google Groups "OpenAPS Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openaps-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openaps-dev/764f09f1-b048-4f6e-a96f-c399bc1f1fa9%40googlegroups.com.

Petr Borodin

unread,
Jan 5, 2020, 7:15:44 AM1/5/20
to OpenAPS Dev
Hi, Scott.
Thanks for your answer. All software versions are up to date. Right now I have not big ability to buy another board for APS.
Unfortunately debugging did not help. Oref0 is stuck here:
'2020/01/05 15:03:00 disconnecting CC111x radio on /dev/spidev0.0
meal.json 35
refreshed: {"carbs":35,"nsCarbs":35,"bwCarbs":0,"journalCarbs":0,"mealCOB":19,"currentDeviation":2.82,"maxDeviation":4.96,"minDeviation":1.96,"slopeFromMaxDeviation":-0.44,"slopeFromMinDeviation":0.223,"allDeviations":[3,2,3,2,5,5],"lastCarbTime":1578220152867,"bwFound":false}
Listening for 10s: .No interfering pump comms detected from other rigs (this is a good thing!)
Continuing oref0-pump-loop at Sun Jan  5 15:04:40 MSK 2020
'
So this operation takes about 2 minutes.
Also there is this operation:
'2020/01/05 15:06:16 disconnecting CC111x radio on /dev/spidev0.0
Allowing wide eventualBG target range: 97.2 - 97.2
Pump profile refreshed; Allowing wide eventualBG target range: 97.2 - 97.2
Settings refreshed; 2020/01/05 15:07:34 connected to CC111x radio on /dev/spidev0.0'

It also takes more than a minute. All other operations are done pretty fast.
Is there anything I can do to speed this up?
To unsubscribe from this group and stop receiving emails from it, send an email to opena...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages