These are chat archives for nightscout/intend-to-bolus

11th
Nov 2018
Eric
@ecc1
Nov 11 2018 00:48
@/all if you run the dev branch and use BLE between your rig and a Dexcom G4 receiver for your BGs and run Debian or Raspbian stretch, please hold off on doing any apt-get upgrades. Something in the most recent Raspbian package upgrades broke BLE comms to the G4 receiver. (Not sure yet if it also happens on Edison rigs.)
cduarte
@cduarte
Nov 11 2018 00:49
I am using version dev 0.7.0!
the autotune is veryyy fast!! :)
Martin Haeberli
@mhaeberli
Nov 11 2018 00:55
@ecc1 thx for heads up. I wonder whether my whacky tethering behavior (on Edisons) is connected. I’ll report back when I know more...
-this article may have cool synergies for OpenAPS users if/when we can figure out how to apply it: https://www.newscientist.com/article/2117854-smartwatches-know-youre-getting-a-cold-days-before-you-feel-ill/
timberFrame
@timberFrame
Nov 11 2018 01:03
I see the effect of a cold in increased insulin resistance before cold symptoms appear. If other measurements also predict a cold, then insulin sensitivity adjustments could be made on that basis.
Eric
@ecc1
Nov 11 2018 01:09
It looks like the BLE issue is Raspberry Pi-specific -- it's the upgrade from version 0.1.8 to version 0.1.9 of the pi-bluetooth package that breaks things (not sure why yet) and reverting fixes it. So if you're affected you can just hold that package rather than all upgrades.
cduarte
@cduarte
Nov 11 2018 01:17
what is the "Days Missing" at the result of autotune?
Scott Leibrand
@scottleibrand
Nov 11 2018 02:14
That’s how many days that hour’s basal was interpolated instead of adjusted (because of meal activity at that time)
Raymond Richmond
@PedanticAvenger
Nov 11 2018 03:45
I'm using IFTTT buttons to log carb consumption and notice an issue on the COB pill on nightscout. Just given the way I have my buttons configured I will use a 50 and a 40 g button to log a 90g meal, but often (not always but often) the calculations only show the 40g in th epill, but hovering shows the full 90 in Careportal COB. It's a bit odd.
Dana Lewis
@danamlewis
Nov 11 2018 04:48
Cob pill is known to be glitch sometimes. Check the OpenAPS pill or OpenAPS logs to confirm cob if needed
RBJervey
@RBJervey
Nov 11 2018 17:25
Hi everyone, is there anyone used to reading papertrail that can look at mine and see if I'm missing something... it looks like OpenAPS gave me a bolus of 6.2 units last night on both nightscout and my pump.. but I can't find it in the logs. (It also super upped by basal rates at that time causing me to crash real hard.... just trying to understand what happened!)
Martin Haeberli
@mhaeberli
Nov 11 2018 17:39

@ecc1

apt-cache policy bluez
bluez:
  Installed: 5.23-2+deb8u1
  Candidate: 5.23-2+deb8u1
  Version table:
 *** 5.23-2+deb8u1 0
        500 http://http.debian.net/debian/ jessie/main i386 Packages
        500 http://security.debian.org/ jessie/updates/main i386 Packages
        100 /var/lib/dpkg/status

(on both working and non-working systems).
but on the broken system:

systemctl status bluetoothd
● bluetoothd.service
   Loaded: not-found (Reason: No such file or directory)
   Active: inactive (dead)

and

systemctl start bluetoothd
Failed to start bluetoothd.service: Unit bluetoothd.service failed to load: No such file or directory.
Eric
@ecc1
Nov 11 2018 17:43
systemd calls it just bluetooth (bluetooth.service), not bluetoothd. 5.23 is probably too old, so you should check that the /lib/systemd/system/bluetooth.service file refers to your /usr/local version of bluetoothd in the ExecStart line
RBJervey
@RBJervey
Nov 11 2018 17:43
@mhaeberli When I run the systemctl and start bluetoothd commands I do them sequentially not together. I don't know that that will helpf with the first error but it's something to try.
Eric
@ecc1
Nov 11 2018 17:44
Also, tab-completion works with systemd ("systemctl status blue<TAB>" etc.)
Martin Haeberli
@mhaeberli
Nov 11 2018 17:58

@ecc1 thx ; looked at both ‘good’ and ‘broken’ rigs; both show in /lib/systemd/system/bluetooth.service that

ExecStart=/usr/lib/bluetooth/bluetoothd

and the md5sums are the same… trying to figure out what’s different ...

Eric
@ecc1
Nov 11 2018 18:00
and on the good rig, what does "systemctl status bluetooth" show? I don't see how it can work if it also is running the 5.23 "system" version of bluez
Martin Haeberli
@mhaeberli
Nov 11 2018 18:03

@ecc1 another hint:
working:

sudo ls -l /proc/745/exe
lrwxrwxrwx 1 root root 0 Nov 10 20:01 /proc/745/exe -> /home/.rootfs/usr/local/bin/bluetoothd

broken:

sudo ls -l /proc/11213/exe
lrwxrwxrwx 1 root root 0 Nov 11 09:49 /proc/11213/exe -> /home/.rootfs/usr/lib/bluetooth/bluetoothd

good system:

systemctl status bluetooth
● bluetooth.service - Bluetooth service
   Loaded: loaded (/lib/systemd/system/bluetooth.service; enabled)
   Active: inactive (dead)
     Docs: man:bluetoothd(8)

about to get clean answer after fresh boot on broken system

looks like somehow the version started on the ‘good system’ is NOT the one referred to in the ExecStart path
broken system after boot:
systemctl status bluetooth
● bluetooth.service - Bluetooth service
   Loaded: loaded (/lib/systemd/system/bluetooth.service; enabled)
   Active: inactive (dead)
     Docs: man:bluetoothd(8)

broken with autocomplete:

systemctl status bluetooth.target 
● bluetooth.target - Bluetooth
   Loaded: loaded (/lib/systemd/system/bluetooth.target; static)
   Active: active since Sun 2018-11-11 10:04:39 PST; 2min 20s ago
     Docs: man:systemd.special(7)

(remote access to reference good system just disconnected…)

Martin Haeberli
@mhaeberli
Nov 11 2018 18:10
@ecc1 so I’m trying to figure out differentially how and where one thing is started in one and not in the other.
@RBJervey can you send an example ? - thanks for your help
Eric
@ecc1
Nov 11 2018 18:12
It looks like the good system must start its own /usr/local/bin/bluetoothd somewhere (maybe from /etc/rc.local, or a @reboot entry in root's crontab?)
But you should get the same effect from editing that bluetooth.service file's ExecStart line
Martin Haeberli
@mhaeberli
Nov 11 2018 18:13
k thx I’ll give that a try
Martin Haeberli
@mhaeberli
Nov 11 2018 18:19
@ecc1 also trying to speculate how installing dev over master broke the failed system...
Eric
@ecc1
Nov 11 2018 18:22
I think you'd have to figure out where the local bluetoothd is getting started on the working rig, because it looks that's no longer happening on the other one
Scott Leibrand
@scottleibrand
Nov 11 2018 18:35
@RBJervey you should be able to find the time of interest, find the pump-loop.log that has the lower IOB before the bolus, and then the one that has the higher IOB afterward, and copy and paste those two runs and anything in between for us to take a look at.
johnnymule
@johnnymule
Nov 11 2018 18:53
My edison has a fully charged battery, but when I unplug it from the charging cable I lose connection. If I restart it unplugged from the charging cable, it works. Am I missing something or does this sound like a board issue?
Eric
@ecc1
Nov 11 2018 19:40
what connection do you lose, specifically? what is in the logs when you disconnect the charge cable?
johnnymule
@johnnymule
Nov 11 2018 20:23
I assumed it was on because the red light stays on, but the papertrail logs stop and I'm not seeing log entries after it gets unplugged from the charging cable until I reset it with the physical button.
Eric
@ecc1
Nov 11 2018 20:27
I'd connect to it via a USB cable and terminal (screen, minicom, etc.) and try to see what's going on
Martin Haeberli
@mhaeberli
Nov 11 2018 20:30
also, note there are TWO red lights. one between the two USB connectors, and one nearer the battery connector. Which one are you talking about. (and is your battery working?)
johnnymule
@johnnymule
Nov 11 2018 21:14
The one between the USB connectors stays on, and the battery will power the rig by itself after a board reset.
Martin Haeberli
@mhaeberli
Nov 11 2018 21:49
ok - I guess I’d go with @ecc1 s advice
Raymond Richmond
@PedanticAvenger
Nov 11 2018 22:43
Is there any variance in the pump battery alert level if you are using lithium vs alkaline or other? i.e. 1.3 correct for lithium?