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

28th
Oct 2017
tepidjuice
@tepidjuice
Oct 28 2017 02:32
Does anyone have an offline i.e. no nightscout, way of entering carbs without bolusing from the pump?
Dave Acklam
@dcacklam
Oct 28 2017 02:33
@tepidjuice IIRC the xdrip interface works without nightscout
As long as you are tethered (wifi or bluetooth)
At least as it's described in the docs, xdrip sends to openaps over the wifi or BT network link, which then sends to nightscout when your rig has internet access again...
@scottleibrand
BTW, one other thing I noticed:
In the event where the radio crashes (either dumps it's firmware or just freezes up), it is very easy to end up in a not-looping condition, because the 'waiting for 40s of silence' function becomes an infinite loop
@scottleibrand This is observed as a never ending line of '.....' after the waiting message, wherin it does not get to mmtune unless you either kill all the bash processes, or reboot.
Thoughts on possible solutions include:
Dave Acklam
@dcacklam
Oct 28 2017 02:40
1) Some sort of timer mechanisim to break out of the 'waiting 40s' if a certain amount of time has happened without 40s of silence being detected
2) Either a function to detect dumped firmware (based on mmtune failing with a '--port not supplied' error), or reflash the radio firmware on each reboot (making the old 'reboot if its not working' thing a way to clear persistant radio errors).
(2) matters because in at least 1 case on my rig (Explorer), I have had to reflash the firmware to prevent the problem described above - even a reboot wouldnt fix it.
I have also noted that reflashing and mmtuning clears the problem - no reboot required.
tepidjuice
@tepidjuice
Oct 28 2017 02:42
@dcacklam as in carbs entered in xdrip+ will be sent by xdripAPS along with your bg to the rig?
tepidjuice
@tepidjuice
Oct 28 2017 03:02
just tested it and it works thanks @dcacklam
Dave Acklam
@dcacklam
Oct 28 2017 04:04
@tepidjuice Also works for insulin boluses. Essentially everything you put into XD goes into OpenAPS
Scott Leibrand
@scottleibrand
Oct 28 2017 06:10
@Bender1061 those changes seemed like improvements, but I'm still testing. Have two flights tomorrow to do some real world stress testing.
@Dcacklam isn't the wait for silence in a for loop that will complete after a bunch of tries?
We haven't had enough radio problems that required ccprog to fix that I would have considered automating that.
Niels Hartvig
@hartvig
Oct 28 2017 09:27
Ideas (I'd be happy to do work on these): Make a small little tool to validate any configuration errors (a simple way to catch syntax errors in json configs). I also thought about making a small nodejs based server that could highlight this as well as an easy way to retrieve log (for those not on papertrail). Only concern would be potential battery impact, but if it was a part of the install script it could lower the bar for those who need to get support later on but are not comfortable with terminal'ing. It could also contain links to troubleshooting (readthedocs, etc).
Niels Hartvig
@hartvig
Oct 28 2017 09:45
(when I say server :arrow_up: - i mean a small webserver so you could point a browser at http://your-rig.local and then see various statuses. I'm aware that both Papertrail and NS support this, but the idea was to help troubleshooting when there's config errors. I was inspired by this from a website "Health Check" feature we built at work which helps non-tech people to see the health of their website and potential ways to solve this - thus limit initial support questions)
PieterGit
@PieterGit
Oct 28 2017 09:50
@tepidjuice : we do a carb entry from the pump, with 0.1 unit of insulin. If in a hypo or going down we will later do carb entry in NS
PieterGit
@PieterGit
Oct 28 2017 09:55
@dcacklam : we should find the root causes of problems instead of doing it the Microsoft way with rebooting and even worse reflashing stuff. I still notice that a /etc/init.d/cron stop ; wait for all loops to finish ; openaps mmtune; /etc/init.d/cron start often is enough to get pump-rig communication back working (without reflashing), so we mush try to find out what's the cause of pump-rig communication not coming back. It might have something to do with the communication state of the pump. In some cases restarting the pump by re-inserting the battery also fixes the problem. Perhaps having an https://en.wikipedia.org/wiki/Exponential_backoff algorithm that will wait eventually longer than e.g. ~ 10 minutes. I believe the pump might recover from miscommunication sessions after some time. On the other hand, we might hit a Medtronic Pump or mmeowlink bug in communication that we are not aware of.
PieterGit
@PieterGit
Oct 28 2017 10:05
@hartvig : i don't think you need a server for that part, you could create a oref0 check comand that implements these checks. This is the way Postfix have implemented this with postfix check. See also: https://github.com/openaps/oref0/pull/759#issuecomment-340147866
tepidjuice
@tepidjuice
Oct 28 2017 10:18
@dcacklam But the pump is my only source of insulin... I guess if I injected with a pen as well might be handy to tip openAPS off about it.
peterlynton
@peterlynton
Oct 28 2017 10:18
@PieterGit Thanks for your help last night! How would I start a PR for new instructions for recreating a git in 0.6.0?
Niels Hartvig
@hartvig
Oct 28 2017 10:21
@PieterGit the post check is a great idea. My idea was mainly focused on lowering the bar for people who "normally" live their life outside of a terminal. So it would be possible to see the health of your rig without the need to type commands. So when people have problems with their rig, the first line of support could be "try to connect to your rig by opening a browser and type your-rig.local".
PieterGit
@PieterGit
Oct 28 2017 10:22
in 0.6.0 the git directory is deprecated, but I think it's not deleted if you upgrade from a 0.5.x stable version. The PR must correct the error on deleting the .git and explain the situation and difference for 0.5.x and 0.6.x. See https://openaps.readthedocs.io/en/master/docs/Resources/troubleshooting.html#dealing-with-a-corrupted-git-repository . I think the git init is mandatory for the 0.5.x release, otherwise openaps commands will fail because it's not a git repo anymore.
peterlynton
@peterlynton
Oct 28 2017 10:36
Sorry a bit lost here - first PR! If I try to add a PR I have to compare across forks, whereas what I think we need is a change to the documentation.
PieterGit
@PieterGit
Oct 28 2017 10:44
@peterlynton I was thinking of a doc change PR only
peterlynton
@peterlynton
Oct 28 2017 10:47
@PieterGit got it - thanks.
PieterGit
@PieterGit
Oct 28 2017 10:54
feel free to ask questions here if making the first PR feels scary, others will help you and your changes will reviewed before being committed. I rather focus on helping users here and improving the software than on making docs PR's.
peterlynton
@peterlynton
Oct 28 2017 10:58
@PieterGit Thanks for your help, PR made.
Nalla306
@Nalla306
Oct 28 2017 11:01
is there a command to check which version of openaps/oref I'm running?
PieterGit
@PieterGit
Oct 28 2017 11:04
@Nalla306 in 0.5.x I think grep -i version ~/src/oref0/package.json should do something like that. 0.6.x will get include a oref0-version or oref0 version in future, see openaps/oref0#759
@peterlynton thanks. I already reviewed it, and let's see if we can raise the bar by adding some extra stuff to your first PR . @danamlewis reviews most/all doc updates when she's around. See https://github.com/openaps/docs/pull/1093#pullrequestreview-72669168
@Nalla306 openaps has openaps --version
Nalla306
@Nalla306
Oct 28 2017 11:07
@PieterGit Thanks!
PieterGit
@PieterGit
Oct 28 2017 11:12
@peterlynton : :thumbsup: for your first PR and modifications, I 'm pretty sure this will be merged quite soon.
kallnap
@kallnap
Oct 28 2017 11:18

ups

edison@AP-Server:~/APS$ openaps --version
openaps 0.1.5

seems outdated

Diadon81
@Diadon81
Oct 28 2017 12:19
Dears, I have a question about pebble and openaps, should I use some special model of pebble watch or I can use any pebble watch?
Scott Leibrand
@scottleibrand
Oct 28 2017 12:36
Any you can still get, since they went out of business.
@hartvig for the offline branch, we could really use a good status web page for displaying information about current BG, IOB, running temp, etc. The main hurdle there is getting a web server running on the rig and basic framework in place for something that can read json files from ~/myopenaps, and then later run commands like oref0-append-local-temptarget.
Dana Lewis
@danamlewis
Oct 28 2017 12:40
@kallnap npm list -g oref0
kallnap
@kallnap
Oct 28 2017 12:42
@danamlewis no need to worry, this is my second oldest rig, that mostly serves for NS hosting and is only used for backup if everything else fails
edison@AP-Server:~/APS$ npm list -g oref0
/usr/lib
└── (empty)

npm ERR! code 1
Dana Lewis
@danamlewis
Oct 28 2017 12:53
openaps is the underlying toolkit - so it's 0.1.5 is unrelated to oref0 version, which should be something like 0.5.5 if you're on current master. Toolkit doesn't have many update and auto update that for you, so only ever need to do npm list -g oref0 to check algo age
fidoman
@efidoman
Oct 28 2017 16:15
My bolus and carb entries are not loading to NS anymore. Getting a 500 error if I try to do it via NS manually. Any ideas? Do I need to donate, backup and clear out the MongoDB?
Dana Lewis
@danamlewis
Oct 28 2017 16:19
What's a 500 error? Sounds like a heroku problem...try restarting dynos
Then I'd check mlab size and try cleanup
Donate/backup/clear out should be absolute last resort, but likely won't be needed
fidoman
@efidoman
Oct 28 2017 16:29
Yes, heroku problem. Already tried restarting dynos no dice. I will try cleanup. Not sure how to do that though.
Dana Lewis
@danamlewis
Oct 28 2017 16:31
K. Checkout the NS troubleshooting page in the docs - shows how to do it
fidoman
@efidoman
Oct 28 2017 16:34
Yep, darn heroku logs say Mongo quota exceeded. great
Martin Haeberli
@mhaeberli
Oct 28 2017 16:38
Screen Shot 2017-10-28 at 09.27.10.png
Screen Shot 2017-10-28 at 09.27.19.png
trying to understand an elevated temp basal decision overnight, in light of temp targets (set to 150-180), but basal decision made in light of an ’89’ target. see enclosed picture:
why 4.5 temp basal just after 8 am?
fidoman
@efidoman
Oct 28 2017 16:41
Is there guidance on what to do when MongoDB quota is exceeded. Their for pay plans are hard for me to understand what my costs would be.
Dana Lewis
@danamlewis
Oct 28 2017 16:42
@mhaeberli eventual bg 161 is why
You might want to turn on OpenAPS forecast, purple lines usually help visualize the future and back up your read of the pill
you do have purple, see it now
But why it's adjusting targets with the high temp target, would want to look st full logs from then
And see how Autosens and adjust targets are interplaying with that
Martin Haeberli
@mhaeberli
Oct 28 2017 16:44
ok, so pull the papertrail logs from plus minus a few minutes ?
Dana Lewis
@danamlewis
Oct 28 2017 16:45
Sure although you may want to go in to both pump loop and Autosens log from those time period
@efidoman I don't know anyone who's needing paid tier; so hopefully it's a one time fluke...no idea what would have barfed and caused the issue, though
But shouldn't be a regular recurrence
Martin Haeberli
@mhaeberli
Oct 28 2017 16:46
btw @danamlewis , on 0.5.5 master on this one...
Dana Lewis
@danamlewis
Oct 28 2017 16:48
@mhaeberli :+1: thx good to know. (We're moving away from advanced target adjustment in dev now; due to other improvements that no longer need that. But check also and let me know if that's turned on in your preferences? That might be what's doing the 102)
fidoman
@efidoman
Oct 28 2017 16:48
@danamlewis - thanks ... I am initiating a new donation of all data. Then I will delete the DB I guess. Not sure what do do there
It seems my entries database table reached a limit of 500mb
I assume I need to clear that out?
Dana Lewis
@danamlewis
Oct 28 2017 16:48
Try cleaning and compresssing first
fidoman
@efidoman
Oct 28 2017 16:48
after donation completes
Dana Lewis
@danamlewis
Oct 28 2017 16:49
Entries shouldn't usually be that big
fidoman
@efidoman
Oct 28 2017 16:49
How do I clean / compress. Didn't see those options on heroku mongodb
Dana Lewis
@danamlewis
Oct 28 2017 16:49
Try compressing - see the NS troubleshooting page in OpenAPS docs
fidoman
@efidoman
Oct 28 2017 16:49
ok
Dana Lewis
@danamlewis
Oct 28 2017 16:49
Bottom end of the left hand nav
Sometimes takes 2-3 tries but will tell you if it's successful and should also see the number go down
Martin Haeberli
@mhaeberli
Oct 28 2017 16:50
per preferences.json : "adv_target_adjustments": true
Dana Lewis
@danamlewis
Oct 28 2017 16:51
@mhaeberli yep, k. Logs next, that's probably the 102. So if you don't like, try tonight with it off
fidoman
@efidoman
Oct 28 2017 16:53
@danamlewis - repair database fixed it. ;-> Thank you! Good side effect is that I also donated a complete updated copy of my data.
Dana Lewis
@danamlewis
Oct 28 2017 16:54
Double yay! Thanks for donating data :smile: :confetti_ball:
I'm not sure 500 error/repair database is on that page; if not, wanna Pr a quick note?
On a ✈️ so maybe is there but don't remember it.
Martin Haeberli
@mhaeberli
Oct 28 2017 17:01
@danamlewis btw, saw this in papertrail logs while trying to find the time window:
Oct 28 08:57:11 etghopenaps0 autosens-loop.log: Setting lastSiteChange to Sat Oct 28 2017 00:38:51 GMT-0700 (PDT) using timestamp 2017-10-28T00:38:51-07:00
Oct 28 08:57:11 etghopenaps0 cgm-loop.log: 148 2017-10-28T08:53:15-07:00
Oct 28 08:57:12 etghopenaps0 autosens-loop.log: ERROR: bad basal schedule [ { i: 0, start: '00:00:00', rate: 0, minutes: 0 } ]
Oct 28 08:57:12 etghopenaps0 autosens-loop.log: +ERROR: bad basal schedule [ { i: 0, start: '00:00:00', rate: 0, minutes: 0 } ]
Oct 28 08:57:12 etghopenaps0 autosens-loop.log: -ERROR: bad basal schedule [ { i: 0, start: '00:00:00', rate: 0, minutes: 0 } ]
Oct 28 08:57:12 etghopenaps0 autosens-loop.log: -ERROR: bad basal schedule [ { i: 0, start: '00:00:00', rate: 0, minutes: 0 } ]
Oct 28 08:57:12 etghopenaps0 autosens-loop.log: +ERROR: bad basal schedule [ { i: 0, start: '00:00:00', rate: 0, minutes: 0 } ]
logs captured (need to excerpt autosens and pump-loop) best way to share afterwards?
fidoman
@efidoman
Oct 28 2017 17:08
@danamlewis - Sure I will PR that if it isn't there
Martin Haeberli
@mhaeberli
Oct 28 2017 17:08
liking the lastSiteChange in the papertrail logs
now just to tie that automatically to the pills ! :-)
Dana Lewis
@danamlewis
Oct 28 2017 17:09
:clap: thx @efidoman
fidoman
@efidoman
Oct 28 2017 17:18
pull request done for 500 error while logging NS treatment - above section about how to repair / cleanup MongoDB
Dana Lewis
@danamlewis
Oct 28 2017 17:20
Ty!
Scott Leibrand
@scottleibrand
Oct 28 2017 17:59
@mhaeberli looks like OpenAPS read a basal schedul with a rate of 0 from your pump, which is not allowed
if you didn't actually set a rate of 0, it might've just been a bad read, and should self-resolve on the next settings refresh
Martin Haeberli
@mhaeberli
Oct 28 2017 18:02
@scottleibrand thx - schedule on the pump AFAIK (as far as I know) has NO zero temp basals - so assuming a bad read ...
Luuc
@luucvp_twitter
Oct 28 2017 23:34
People, in Europe the time will go back to 1 hour tonight. Do I have to change something on my Edison or will it handle the shift just fine?