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

15th
Sep 2018
boc-the-git
@boc-the-git
Sep 15 2018 00:14

With further investigation into my issues uploading entries (BG data) to NS, I've come across the following:
2018/09/15 09:56:18 POST https://redacted.herokuapp.com/api/v1/entries
2018/09/15 09:56:18 {
"type": "sgv",
"date": 1536861480000,
"dateString": "2018-09-14T03:58:00+10:00",
"device": "openaps://Liverbird",
"sgv": 86
}
2018/09/15 09:56:18 {
"status": 401,
"message": "Unauthorized",
"description": "Invalid/Missing"
}

I ran the following to trigger that output: cgmupdate -v -f monitor/cgm-mm-glucosedirty.json -u -b 30h -k 48h

My NS is configured to use token auth and the rig is successfully connecting to it as treatment and devicestatus records are getting uploaded successfully. (Running oref 0.7.0-dev, with MDT cgm and a 522 pump. )

The following gives no results, might be relevant..?
root@Liverbird:~/myopenaps# grep "MDT cgm" openaps.ini
root@Liverbird:~/myopenaps#

Raymond Richmond
@PedanticAvenger
Sep 15 2018 03:05
Did anyone else online get one of those edison 5 packs from E-bay? The yocto image isn't (apparently) using the default root pw and I've been waiting on the seller for a bit and hoping to shorten that wait. ;)
Eric
@ecc1
Sep 15 2018 04:07
I wouldn't use any computer from eBay without wiping it first, and that goes for Edisons too. I'd reflash with a known system.
@BOC14 token auth isn't supported yet by the go uploading code. It's being worked on though (not by me, but I plan to merge it as soon as I can)
lyntonr
@lyntonr
Sep 15 2018 04:36
I dont know why but my autotune results are way way off . It is recommending very low carb ratios which I suspect is causing hypos. so I need to disable it. If I just remove the line "enable": " autotune ", line from my preferences file will that do it ?
lyntonr
@lyntonr
Sep 15 2018 04:47

heres output from running one days worth of data as a standalone test,

Limiting CSF to 8.58 (which is 1.2 * pump CSF of 7.147058823529412 )
totalMealCarbs: 82 totalDeviations: 1028.2 oldCSF 8.241 fullNewCSF: 12.54 newCSF: 8.576
Limiting fullNewCR from 2.429 to 4.76 (which is 0.7 * pump CR of 6.8 )
oldCR: 4.779 fullNewCR: 4.76 newCR: 4.775
Limiting adjusted ISF of 31.74 to 40.50 (which is pump ISF of 48.6 / 1.2 )
Limiting ISF of 39.61 to 40.50 (which is pump ISF of 48.6 / 1.2 )
p50deviation: 0.355 p50BGI -1.88 p50ratios: 0.806 Old ISF: 39.383 fullNewISF: 31.743 adjustedISF: 40.5 newISF: 40.5 newDIA: 6 newPeak: 75

Autotune pump profile recommendations:
---------------------------------------------------------
Recommendations Log File: /root/newdirectory/autotune/autotune_recommendations.log

/usr/local/bin/oref0-autotune-recommends-report: line 25: warning: setlocale: LC_NUMERIC: cannot change locale (en_US.UTF-8): No such file or directory
Parameter      | Pump     | Autotune
-------------------------------------
ISF [mg/dL/U]  | 48.600   | 40.500
Carb Ratio[g/U]| 6.800    | 4.775

If anyone can see anything suspicious here please comment. Im not really sure what all this means.

boc-the-git
@boc-the-git
Sep 15 2018 04:59
Thanks for the info @ecc1. FYI @danamlewis, sounds like using token based auth is my issue for NS uploading. Tomorrow or Monday I'll look to change my auth and test that.
Dana Lewis
@danamlewis
Sep 15 2018 05:10
Ahhh. Thanks for heads up
Scott Leibrand
@scottleibrand
Sep 15 2018 07:48
@GlennCoco check which node process it is. It might be autosens running more often than necessary due to the full pumphistory refresh we started doing after every loop to fix a race condition.
If so we might want to change it to run much less often.
Scott Leibrand
@scottleibrand
Sep 15 2018 08:05
@lyntonr the earlier output should show which time periods were being used to calculate CR, marked off by CR start time and end time.
lyntonr
@lyntonr
Sep 15 2018 09:12

Thanks for chipping in @scottleibrand here is a gist of the whole output. One bit that looks odd (to me) says :

Ignoring 459 m CR period.

https://gist.github.com/lyntonr/83bbb806eb385762ba538a1770256b62

GlennCoco
@GlennCoco
Sep 15 2018 09:28
@scottleibrand I am not sure, running ‘top’ command and under the command column it just says node.
fidoman
@efidoman
Sep 15 2018 11:04
My history of insulin delivery shown in Nightscout is incorrect yesterday for everything before 12am last night. Anyone know if this this an openaps or NS issue? See screenshot below.
Screenshot from 2018-09-15 07-05-53.png
Eric
@ecc1
Sep 15 2018 12:14
Do you have the pump history corresponding to the incorrect period? Also, what rig and openaps branch are you using?
alimhassam
@alimhassam
Sep 15 2018 12:18
@efidoman you can also have a look at your ns treatments reports to see if that gives a hint on what the issue might be?
we've seen a similar issue long time ago, but that was fixed, not sure if its' the same thing.
fidoman
@efidoman
Sep 15 2018 12:49
Edison rig and my branch is "before-refactor" so I may be missing the fix.
Just checked nightscout again and all the boluses went back to normal without me doing a thing. Weird.
alimhassam
@alimhassam
Sep 15 2018 13:17
yeah, things used to get back to normal after the full daily refresh that occurs every 6 hours.
very likely it's the same bug we've seen before.
it shuold be fixed by using a newer openaps.jq
Jens Heuschkel
@juehv
Sep 15 2018 13:18
@BOC14 sorry for the late response!
you got my suggestion right ... but it's obsolete if the token auth was the problem
is you issue fixed by now ?
David
@davidkeddydb
Sep 15 2018 15:06
@ecc1 @danamlewis Thanks. I guess if I am that committed to going back to Omni in the short run, I may have to ditch my iphone or get a second phone for AAPS.
Scott Leibrand
@scottleibrand
Sep 15 2018 19:06
@glenncoco google for how to identify a node process. If nothing else you can get the process ID with top and get more details with ps.