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

26th
Sep 2017
Aaron Michelson
@Aaybob
Sep 26 2017 02:57
@darrylschick thanks, I guess I needed a little encouragement to update...thanks
dr-crash
@dr-crash
Sep 26 2017 03:10
I hope someone can help me... This morning I woke up to find my Edison quit on me sometime last night. I was getting CGM and treatment data but no device status updates (like edison or pump battery levels) or OpenAPS adjustments. I've since tried reinstalling OpenAPS, but now nothing gets through. The log shows some kind of error with Preflight and mmtune:
Starting supermicrobolus pump-loop at Mon Sep 25 22:03:02 CDT 2017 with 2 second wait_for_silence:
MDT CGM configured; not waiting
Radio ok. Listening: .No pump comms detected from other rigs
Preflight git.exc.InvalidGitRepositoryError: /root/myopenaps
fail. SMB pump-loop failed. pump_loop_completed more than 15m old; waiting for 30s silence before mmtuning
Radio ok. Listening: .No pump comms detected from other rigs
Listening for 30s silence before mmtuning: .No pump comms detected from other rigs
mmtune: git.exc.InvalidGitRepositoryError: /root/myopenaps
Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
Unsuccessful supermicrobolus pump-loop at Mon Sep 25 22:04:24 CDT 2017
dr-crash
@dr-crash
Sep 26 2017 03:17
Running oref0-reset-git returned: Attempting to fix git corruption. Please wait 15s.
Git repo does not appear to be corrupt.
/usr/local/bin/oref0-reset-git: line 55: 28145 Terminated ( sleep 15; killall oref0-fix-git-corruption )
find: `/tmp/git-1506395625': No such file or directory
then tried rm -rf ~/myopenaps/.git but the log shows the same errors
dr-crash
@dr-crash
Sep 26 2017 03:47
looks like that followed by rerunning setup and disabling SMB's restored the CGM and treatment data on nightscout. Hopefully looping will pick up in a little while
dr-crash
@dr-crash
Sep 26 2017 03:55
there it goes! thanks Dana :)
Dana Lewis
@danamlewis
Sep 26 2017 04:01
:+1:
br
@brmccollum
Sep 26 2017 04:43

@danamlewis purchased, will cross my fingers. I tested several different brand new batteries over the past few days. Cheap batteries will give very poor readings ("916.492", 1, -91) inconsistently always lower than the default 916.363, more expensive batteries like "Energizer Ultimate Lithium" never once has produced anything other than a -99. I feel like it is likely the pretty sensitive to the power coming into it on the 712.

I did have one more question before I attempt to get the 715 working:
Do I need to delete the extra files that I created for the 712 before inputting the 715 SN?

Dana Lewis
@danamlewis
Sep 26 2017 04:44
oh fascinating - thanks for sharing the battery results.
and nope - just swap the 715 SN into your runagain and run it again - it'll overwrite the x12 files automatically
br
@brmccollum
Sep 26 2017 04:44
Great, thanks :+1:
Dana Lewis
@danamlewis
Sep 26 2017 04:45
yea, bummer re your 12s.
br
@brmccollum
Sep 26 2017 04:45
Yeah, if I get the 715 working, I would be glad to send them to anyone who wants to mess with them
Niels Hartvig
@hartvig
Sep 26 2017 04:52
@scottleibrand yes, I use MDT
LilDucky
@LilDucky
Sep 26 2017 05:07
@brmccollum I am extremely interested in the pumps for testing
Tim Street
@tim2000s
Sep 26 2017 05:27
Looks like there’s more reason to not use lithium’s in the 712 than we thought. I’ve a 715 that runs okay on Energizer Lithiums though.
Scott Leibrand
@scottleibrand
Sep 26 2017 05:34
@hartvig I believe that is a problem with the MDT CGM setup that needs someone to look into and figure out why. apparently it doesn't break looping, though, or someone would have by now
LilDucky
@LilDucky
Sep 26 2017 05:39
@scottleibrand I have seen that unzoned error too, as far as I recall mainly after comms losses. will do a log trawl and see what I can find
Luuc
@luucvp_twitter
Sep 26 2017 08:02
So, I'm currently on the Medtronic sensor in combination with Veo pump on openAPS. Often it is not very accurate, so I'm looking into the dexcom sensors. For openAPS which one would you rather recommend; G4 or G5?
Niels Hartvig
@hartvig
Sep 26 2017 09:09
@scottleibrand I'd be happy to investigate and see if I can find a root cause too! :thumbsup:
mkericson
@mkericson
Sep 26 2017 12:48
@luucvp_twitter will work with either. Go with G5 for BT and you will not need to carry the receiver.
After troubleshooting offline looping for a while with no success I decided to get a new data cable. I purchased this one first :: https://www.amazon.com/gp/product/B01M5GZ3N0/ref=s9u_simh_gw_i3?ie=UTF8&fpl=fresh&pd_rd_i=B01M5GZ3N0&pd_rd_r=8ZA35ZH2HRN5XB4S5P50&pd_rd_w=KDeWZ&pd_rd_wg=zAWFk&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=&pf_rd_r=DFEN10QWAZXCCGCH3FFZ&pf_rd_t=36701&pf_rd_p=1cf9d009-399c-49e1-901a-7b8786e59436&pf_rd_i=desktop
Then I purchased that exact one in the docs and now its working.
LilDucky
@LilDucky
Sep 26 2017 12:52

@scottleibrand the error occurs every loop just after it enacts.

glucose.json newer than pumphistory. : Refreshed pumphistoryWarning: clock input Invalid Date is unzoned; please pass clock-zoned.json instead and meal.json

Running Medtronic 754 and Medtronic CGM with 0.5.4 on an Edison rig.
glenrob1940
@glenrob1940
Sep 26 2017 13:26
Are there concerns with bluetooth hacking linux devices using openaps? The current advice for linux bluetooth in general appears to be to turn it off when not needed because of hacking issues. We, of course, can't turn it off when running openaps.
Paul Andrel
@stavlor
Sep 26 2017 13:45
@scottleibrand was doing some testing yesterday with determine-basal off of dev, anything major change in supplied inputs recently was seeing a good bit of nulls in UAM PredBG and EventualBG was getting NaN
openaps":{"suggested":{"temp":"absolute","bg":152,"tick":"+5","eventualBG":null,"snoozeBG":147,"insulinReq":null,"deliverAt":"2017-09-25T21:24:27.123Z","minPredBG":131,"predBGs":{"IOB":[152,155,158,160,161,162,162,162,161,159,157,154,150,147,144,141,139,137,136,134,133,132,132,131],"UAM":[152,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null]},"COB":0,"IOB":2.48,"reason":"COB: 0, Dev: 77, BGI: -8.61, ISF: 35, Target: 100, minPredBG 131, minGuardBG 999, IOBpredBG 131; Eventual BG NaN >= 100,  insulinReq NaN; setting 30m low temp of 0.12U/h. ","rate":0.12,"duration":30,"timestamp":"2017-09-25T21:24:27Z"},"iob":{"iob":2.48,"basaliob":-0.006,"activity":0.0492,"time":"2017-09-25T21:24:27Z"}},"uploaderBattery":100,"created_at":"2017-09-25T21:24:27Z","NSCLIENT_ID":1506374667218},
Roumen Georgiev
@PoweRGbg
Sep 26 2017 13:48
Look for meal_data.maxDeviationSlope and meal_data.minDeviationSlope . I had the same problem few mins ago and adding these fixed it
jaylagorio
@jaylagorio
Sep 26 2017 13:54
@glenrob1940 As far as I know BlueZ isn't patched yet so it's going to take at least until they get that done to get it updated on Edison/Pis.
lemoir
@lemoir
Sep 26 2017 14:29
Hi everyone, not sure anyone can help, I'm looking at building an APS system for a friend of mine. She has a Medtronic pump already but no CGM. Whats the best way of obtaining something like a G4 for a reasonable price? Everything will be self funded and there isn't much money available ( a very common tale I'm sure )
Eric
@ecc1
Sep 26 2017 14:58
People who have upgraded to G5 might offer their G4 receivers at reasonable prices, or free. But be aware that the real cost will be the sensors.
lemoir
@lemoir
Sep 26 2017 15:15
yes I'm aware that that is the case, but yes I was hoping that might be the case for G4s. Where should I look though?
Eric
@ecc1
Sep 26 2017 15:35
You could post a request on the various Facebook groups (CGM in the cloud and Dexcom users) if it's OK with their moderators, and there are various T1D "pay it forward" groups too.
Scott Leibrand
@scottleibrand
Sep 26 2017 15:49
You can also look into using xDrip instead of a receiver, and possibly a transmitter that someone has replaced the batteries on. Both techniques are widely used in the UK where lots of people self-fund CGM.
lemoir
@lemoir
Sep 26 2017 16:20
thanks a lot for your advice I'll see what I can find
cameronrenwick
@cameronrenwick
Sep 26 2017 16:55
is there any way to use an xdrip+ diy receiver to direct connect to one's rig for offline looping much the same as using an actual dexcom receiver to connect and loop?
Diabetic_Dude
@diabetic_dude_twitter
Sep 26 2017 17:49
@cameronrenwick wouldn't that be nice..... However, I think that all require the NS nite (which means online) aside from the offline solution mentioned in the docs (plugging dexcom directly into your rig) if you find something cool please mention me because I would love to use it also.
Dana Lewis
@danamlewis
Sep 26 2017 17:54
@cameronrenwick @diabetic_dude_twitter a few people have looked into that but not sure anyone's gotten it going
@luucvp_twittter many of us are very happy with G4 over G5 because plugging in receiver gives us raw data. G5 doesn't have raw data. So I'd factor that into the decision.
sdneufer
@sdneufer
Sep 26 2017 17:55
@cameronrenwick @diabetic_dude_twitter You need the xdrip hardware/software to be re-designed to send glucose data through a usb connection that can be read on the edison like the receiver is. I have thought about this but not enough to see if doable.
Savek-CC
@savek-cc
Sep 26 2017 17:59
@sdneufer I think that can already be done - xdrip hw with continous 5min sensor data via USB...
sdneufer
@sdneufer
Sep 26 2017 18:02
@savek-cc Does not xdrip+ software do some manipulation on the data before it sends to xdripAPS or RESTAPI... That manipulation would have to happen somewhere.
Dana Lewis
@danamlewis
Sep 26 2017 18:06
Another option is Bluetooth
sdneufer
@sdneufer
Sep 26 2017 18:10
If the data that the uploader rig sends to the Android phone running xdrip+ is in the form the edison needs then one could cut out the Android phone. I have not looked at that data to see how different it is from what is needed.
@danamlewis That is the bluetooth option from uploader xdrip rig to edison I assume you are talking about.
jaylagorio
@jaylagorio
Sep 26 2017 18:13
Or there would be a processing script on the Edison that puts the data in the form Nightscout needs
sdneufer
@sdneufer
Sep 26 2017 18:15
I think openaps would upload it if the processing script made it available and the edison was online. Ehh?
Dana Lewis
@danamlewis
Sep 26 2017 18:15
Yep to what I meant, and yep
jaylagorio
@jaylagorio
Sep 26 2017 18:16
In cases where the Edison was not online you'd probably want to store the data for upload to Nightscout until such time that it is. The transmitter is very fire-and-forget, and you can't query back in time like you can with the receiver.
Niels Hartvig
@hartvig
Sep 26 2017 18:31
What's the purpose of clock.json vs clock-zoned.json? Trying to get to the bottom of the zoned error messages in the log when using MDT
cameronrenwick
@cameronrenwick
Sep 26 2017 18:31
sigh...thanks everyone... these are exactly the thoughts that have been running thru my head...
sdneufer
@sdneufer
Sep 26 2017 18:32
So now that everybody's head is on the same page, who can do what about it?
kallnap
@kallnap
Sep 26 2017 18:37
isn´t the xdrip-js project working on getting BG data via Bt directly from G5 transmitter to the edison?
sdneufer
@sdneufer
Sep 26 2017 18:38
Is there a G4 uploader rig to Edison via bluetooth project also?
jaylagorio
@jaylagorio
Sep 26 2017 18:40
@sdneufer I think there was some python that existed that purported to do that very early on, but I was never able to get it working and I don't know of anyone else that got it working after me. The project looks so different now (with the setup script and everything else) I don't even know if that code hung around
Dana Lewis
@danamlewis
Sep 26 2017 18:42
There was openxshareble for G4 receiver -> rig, that no one has touched in over a year
To @kallnap point, there's xdrip-js group working on Bluetooth to rig from G5 transmitter
jaylagorio
@jaylagorio
Sep 26 2017 18:44
Yes, openxshareble was the thing I tried and never got working, other comments at the time was that it was flaky at best
kallnap
@kallnap
Sep 26 2017 18:44
@danamlewis g4 share only worked for US as far as I know
@oschumac had soldered a wixel to the edison mini breakout board and got that working as G4 receiver, but I never got that far and I am not aware that anyone ever connected the xbridge hadrware via BT to the edison, although that would have been nice
kallnap
@kallnap
Sep 26 2017 18:56
I am currently using xdrip+ and the offline loop option mentioned in the docs, which works fine although BT hotspot is quite battery heavy
Scott Leibrand
@scottleibrand
Sep 26 2017 19:00
@hartvig clock-zoned.json should have timezones like +0200 on the timestamp. the raw clock output from the pump is unzoned, so we have to add the rig's timezone.
kallnap
@kallnap
Sep 26 2017 19:02
@sdneufer unfortunately there seems to be no perfect solution for offline looping. I did run NS and mongoDB for a while on my edison for best offline loop experience, but then NS changed so much that I can no longer run it on my edison
tazitoo
@tazitoo
Sep 26 2017 20:15
@danamlewis - openxshareble still works - I am in permanent camping mode with my rig and it's reliable - for me. I just noticed the jubilinux update, I'll have to get that installed on a spare edison to check what differences there are with bluetoothd (other than new systemd crap...)
Dana Lewis
@danamlewis
Sep 26 2017 20:16
@tazitoo you're not using any of the latest oref0 though are you? If I remember correctly, you had a variety of workarounds?
I think you had a repo w/ some notes about it at one point, can you share here again for those interested in poking at it? Thanks!
Scott Leibrand
@scottleibrand
Sep 26 2017 20:17
if you (or anyone) wants to take on the project of integrating openxshareble with oref0-setup and testing that it all works on a freshly installed rig, that could be a big win
Dana Lewis
@danamlewis
Sep 26 2017 20:17
:+1:
Scott Leibrand
@scottleibrand
Sep 26 2017 20:18
as with bluetooth tethering, getting anything BLE to work reliably may simply be a matter of standardizing the setup to something that works and fully automating that so no one deviates from it
tazitoo
@tazitoo
Sep 26 2017 20:19
I'm pretty up to date with oref0 - I have some steps I used to reduce the amount of communication from rig to dexcom receiver and pump - and if that's the real bottle neck for reliability, any thoughts about implementing local DB to hold previous readings? Pulling 24 hrs of data every 3 minutes seems like a tall requirement...
...having a local DB is more on my mind as I wonder about using autotune and smb without using Nightscout
Scott Leibrand
@scottleibrand
Sep 26 2017 20:20
we did a bunch of that sort of stuff in the myopenaps-cgm-loop instance: getting a 24h json file and a short recent one and merging them...
tazitoo
@tazitoo
Sep 26 2017 20:21
@scottleibrand - does that help with having data in hand for autotune?
Scott Leibrand
@scottleibrand
Sep 26 2017 20:22
autotune can also work on local files (the first prototypes only did that), but we found it was easier to integrate with NS so that's what we automated
it might require some debugging, but it shouldn't be too heavy a lift to get autotune-prep working with local files again and fully integrate it as an option to oref0-autotune.sh
we already have 24h of pumphistory and 24h of glucose data on hand, which is almost enough for a full nightly autotune run
tazitoo
@tazitoo
Sep 26 2017 20:25
thanks for the pointers. I'll work on getting the new edison set up with jubilinux v0.2.0 (...whatever the latest stable release is...) and document my steps (especially w.r.t oref0-setup.sh). Then I'll check on running autotune from local json files.
Scott Leibrand
@scottleibrand
Sep 26 2017 20:25
we might actually want to pull about 30h in order to have an accurate picture of insulin activity at the 24h-ago mark, but that'd be a pretty easy change
or maybe we already do that for autosens, I can't remember
Dana Lewis
@danamlewis
Sep 26 2017 20:26
:+1: thanks for jumping in @tazitoo !
tazitoo
@tazitoo
Sep 26 2017 20:27
np - just paying it backwards. :+1:
Dana Lewis
@danamlewis
Sep 26 2017 20:27
:D :D :D :clap:
Scott Leibrand
@scottleibrand
Sep 26 2017 20:27
sideways? ;-)
Dana Lewis
@danamlewis
Sep 26 2017 20:27
all the ways! :)
tazitoo
@tazitoo
Sep 26 2017 20:28
you two ... <sigh> :smile:
garykidd
@garykidd
Sep 26 2017 21:17
Not sure why I keep getting password notice xhangrson dexcom when I'm not changing password. Hmm
daveklingler
@daveklingler
Sep 26 2017 21:31
Can anybody tell me what the nominal behavior is for the LEDs on the Edison/Explorer combo? Does the power light normally stay on all the time? I think I've caught it flickering a couple of times. @scottleibrand ?
dan lightner
@danimaniac
Sep 26 2017 21:37
@daveklingler The Explorer Board has 3 led's. Power led is always on (assuming, of course, the board is powered on :) ). Charge led is only on when the connected battery is charging, not on if fully charged. Radio led on when the the board is communicating with the pump (very brief flicker).
Tim Street
@tim2000s
Sep 26 2017 21:48
Hmm. Interesting bluetooth issue. The bluetooth daemon runs quite happily and is discoverable, but when trying to pair, only the phone creates the passkey. It is never shown in bluetoothctl. So it's impossible to pair. I've restarted the edison, but no luck. I've also restarted the entire daemon and HCI config.
This is running on Jubilinux 0.2.0 with BlueZ 5.44
Anyone else seen similar?
It errors out with incorrect PIN or passkey
(on the phone)
Scott Leibrand
@scottleibrand
Sep 26 2017 21:54
@daveklingler @danimaniac the two radio LEDs next to the microUSB ports briefly light up when the cc1110 is being reset, which we do every time we mmtune.
jaylagorio
@jaylagorio
Sep 26 2017 21:54
@tim2000s Are you telling the phone to pair with the Edison or the Edison to pair with the phone?
Also what phone/OS version?
Tim Street
@tim2000s
Sep 26 2017 21:56
Android 8.0 - happily paired with a different rig, so it's asking the phone to pair with the rig.
jaylagorio
@jaylagorio
Sep 26 2017 21:57
I had my Edison initiate the pair and didn't run into that problem, want to try it from that angle?
Wait, are the two rigs named the same? I bet it's confusing the phone if so. You'll want to change the rig's BT name.
I ran into that with my dev rig
Tim Street
@tim2000s
Sep 26 2017 22:45
All rigs are differently and no I can’t connect rig to phone either.
I’ve had the rig connect to one phone only so far. I normally have two configured in bluetoothctl The other behaves oddly by not auto connecting, although you can manually get a by link which is very solid.
Zakm0n
@Zakm0n
Sep 26 2017 23:33
I'm having some major rig issues today. my git corrupted, and I followed the troubleshooting at http://openaps.readthedocs.io/en/latest/docs/Resources/troubleshooting.html?highlight=troubleshooting%20git
but now i'm getting this: ep 26 18:32:42 Aps pump-loop.log: Preflight git.exc.InvalidGitRepositoryError: /root/myopenaps
Sep 26 18:32:48 Aps pump-loop.log: fail. Preflight git.exc.InvalidGitRepositoryError: /root/myopenaps
Sep 26 18:32:48 Aps pump-loop.log: fail. SMB pump-loop failed. pump_loop_completed more than 15m old; waiting for 40s silence before mmtuning