@scottleibrand I think I will have the Edison flashed in the am as I was able to follow @diabeticgonewild example of flashing the card using cgywin. If I use the Oref0 dev set up, as I will be flashing Jubilinux, should I just set it up without mmeowlink initially and just change it after the fact? I think based on the instructions that is the easier way to go.
oref0-setup will install mmeowlink if it's not present
Cool. I'm at the exact same spot.
thanks, @diabeticgonewild . Total lifesaver.
ok cool so I wont have to put all the tty stuff in as it wont be set yet
you will want to tell oref0-setup which TTY to use, or it will assume you are using a carelink
I think the /dev/ttyACM0 or whatever will show up even before you install mmeowlink
Was thinking it might be cool to plot basal IOB on NS---got to imagine thats not too difficult--or has someone done that?
my Anker, despite promise in the Amazon FAQs, does not actually do passthrough. Does anyone have an amazon link to a real battery pack that works for them and their Edison rig? Or, should I suck it up and try a polymer lithium ion? If I lose power to my Edison, is it as dire as it was on the Pi?
@scottleibrand Great. I will be going off the OTG to the TI. Will see what happens tomorrow.
@lollydaggle, @cjo20 told me other day that the little 3350mah battery I have with what we are going to run should do 13 hrs or so pretty easily
signing off. Thanks Monica @diabeticgonewild for directing me to the youtube video. Really helpful. Hopefully will have it flashed tomorrow am.
@lollydaggle the Edison seems to handle power-off much better than the Pi.
@davidkeddydb @lollydaggle you're welcome
@garykidd there’s already an option to visualize basals in NS, or are you talking about something different? https://twitter.com/danamlewis/status/760839954012381186 example of blue line at top is basal visual
@danamlewis yeah I have that basal line but what I was thinking of was not basal being delivered but tracking--the basal IOB....you can kind of see it with the basal delivery but not really. Another thing would be to see what the basal delivery rate was per hour, or on a rolling 2, 3, or 4 hour basis. It's all good stuff...I think a lot of that data may be able to help set and/or tailor your base basal rates.
would you want just basal iob, though, or netiob impact visualized? thinking the latter might be more useful.
I agree for per hour is useful looking back - I think we talked at one point about making that a NS report but never saw if that got completed, but I know Milos & Jason were adding to the reporting capabilities so might check & see what they’ve added there
regarding the net IOB is that just taking into account basal IOB and any bolus derived IOB? Still being able to see visualize the basal IOB (whether positive or negative) may be pretty insightful into establishing settings and being able to better visualize the effect of the last [4] hours of basal adjustments.
are you planning to use it real time or retrospectively?
real time vis is one thing, retrospective reporting and infrequent usage would be a different development scenario
@danamlewis I love your NS pic though...hope to get closer to that soon.
netIOB is both basal and bolus impact
That's an interesting thought
@tim2000s
I had wondered if reason companies were getting FDA approval way sooner than expected, might also be an attempt at regulating open access. (Like black market or something)
No. They're all aware of a limited supply of old loopable pumps out there.
I think the writing has been on the wall re loop
Looo
*looping being safer than not looping. And Medtronic did the required studies, but FDA didn't need a long time to figure out it was ok to approve
Writing on the wall---thanks to you guys! 👍👍❤️
Lots of academic studies all showing the same thing, too.
Ya. But I agree the collective voices of our community is probably contributing in a positive way, too :)
@danamlewis probably both real time visualization and maybe a report to look at historical data better. I think being able to click the 24hr button and see not only a blue line for actual basal being delivered but let's say a red line that would show basal iob would be useful. You'd be able to see straight away if your basal iob may be impacting treatment for example.
Was talking something similar the other day with @sulkaharo , but around visualizing impact of any iob
If someone's going to add another real time vis, combined net iob impact would be my vote
And for retrospective, some of the reports to get a feel for any needed basal changes.
Or really, what we need to do is oref0 issue 99, a report to tell you needed basal changes and other setting changes :)
moonshot
@scottleibrand when running oref0-setup.sh after I enter all my information in the setup it gets stuck.
Continue? y/[N] y
Checking /home/edison/johnopenaps: Initialized empty Git repository in /home/edison/johnopenaps/.git/
[master (root-commit) b3de1ec] initializing openaps 0.1.6-dev
1 file changed, 0 insertions(+), 0 deletions(-)
create mode 100644 openaps.ini
/home/edison/johnopenaps initialized
/home/edison/src/oref0/ already exists; pulling latest
Already up-to-date.
Checking oref0 installation
it just sits there
that is on oref0-setup branch
I thought I fixed that in oref0-setup branch.
Try the commands it's running at that step and see which one hangs.
I somehow flashed the edison, can't get wifi to work yet. I still have no clue on the battery and I don't know why its so difficult. If I use a polymer lithium ion, I need a sparkfun battery board, correct? Which are on back order.
unless there's some magical battery pack that people are using
or if I can buy multiple battery packs, keeping one always charged and swap them out
Soldering to power block would be the only other option. Or a battery block. Or in a month use the explorer board :)
@lollydaggle adafruit.com carries LiPo boards as well. Just make sure if you're buying batteries and charging circuit boards that the polarity matches (+/-). I've got a few batteries that are backwards to my charging board.
@scottleibrand @danamlewis The setup script is amazing, thanks for putting it out there! It was the impetus I needed to update to AMA. Question is, since getting a new rig running with the setup script (using AMA, everything else is default besides max iob and max basal rates, which I changed), I've had two Azure quota-exceeded outages in the last week. I've used Nightscout for 1.5 years, OpenAPS for 5 months now and it's never happened until now. Ideas on what I might do (besides hosting my own mongo/webapp on a cloud-server)?
Just FYI, it's absolutely hilarious to watch COPS on TV when high on THC/marijuana...I remember the first time the Marinol high hit me, and I happened to watch TV and COPS was on. It never gets old!!!!
You could help us use the NS-device as part of the script. Or swap to heroku in the meantime.
I'm blazed right now. Been nauseated lately a lot and don't have school so...might as well be high...
We need help optimizing for low data usage if anyone feels strongly about sticking to azure. Scott updated the docs re heroku but may need to make it more prominent
Sorry, I've only been paying half-attention to gitter the last couple weeks, so I'm not sure what you mean by "use the NS-device as part of the script"?
Heh. There's a NS-device that people have been using manually, but is not part of the setup script (yet, until someone helps with that part)
Got ya. I'll dig in Gitter for the NS-device conversations, and probably quick toss-up a Heroku server. Thank you, @danamlewis .
:+1:
Note to @scottleibrand re pulling the wiki and pointing people to dev docs for now since it has the heroku and other notes
With all the extra stuff everyone wants (like BG uploads, rawbg, etc.) I'm not sure we'll ever get back down below Azure free quotas. But if anyone wants to try optimizing the loop for that, replacing my custom aliases with the proper calls to the
ns device is the place to start.
Yeah @garykidd / @danamlewis we could plot a number that was based on expected insulin impact, taking the absorption time into account. IOB on it's own is slightly misleading -> right after a bolus IOB = bolus, but we all know the real impact will only happen once it absorbs. Basal is similar - IOB is very good at modeling the end result of the currently injected insulin but not good at modeling when this will happen.
Yay just had to puke. 5th time in 2 days...way to go, gastroparesis.
@scottleibrand Understood, and I'm certainly not complaining. All things AMA and the setup script are treating me very well! With the amount of sleep I get right (2 month old isn't in a great rhythm yet), I might just pay $7/month for a working solution to keep NS up and running. FYI, only input I had on the setup script was that the wording on entering the device name for mmeowlink confused me: "/dev/ttySOMETHING" made me think that I was supposed to enter "ACM0", instead of "/dev/ttyACM0". Completely my oversight, but it might be clearer if the verbiage in the step-by-step of the script said something like "What you enter will probably look like "/dev/ttySOMETHING", without the quotes".
That sounds like an excellent thing to PR in. :smile:
+1
What, you mean my comment about puking?
Git doesn't do Barf Requests yet, can you hold it in until the next version comes out?
+1 for a PR by @mikestebbins
LMAO!
LOL @sulkaharo
Hopefully you know I was only joking about the PR...
That was awesome @sulkaharo
Hahaha! I am watching COPS right now and a guy the cops detained "has insulin syringes in his car because he has borderline type 2 diabetes" hahaha
At about 14:00 https://youtu.be/kH1wAClVUsw
@scottleibrand The 640G allows manual calibration of the sensor, so I'd expect that from the 670 as well.
@danamlewis I've flipped my NS site to heroku and it doesn't suffer the Azure lock out on the $7 plan. It's also estimating that the monthly cost won't exceed the $7. For my money, it's worth swapping over for the certainty.
:+1:
(But if you're swapping to Heroku for an EU based G5 with Share the instructions are missing some setup details. I'll try and add them later today).
Heroku did not let me set up my page, apparently my e-mail provider was on a black list (web.de)
Oops!
Time for new email address!
@tim2000s I just recently went offline with my loop, so I hope I do not need to switch to Heroku
My issue is Dexcom G5. Can't take that offline without moving to Loop. Which is why I want to port oref0 into Loopkit.
@tim2000s does Xdrip/Xdrip+ in the meanwhile fully supports G5? if so you could go with my poor mans solution and install mongoDB and NS on your Edison. I have my cell run as a hotspot. The only issue there is the limited space on the Edison. My DB has 260 mb after roughly 2 weeks of looping. This is wihtout Autosense and AMA and therefore not so many meal entries.
@C-Ville I'd rather mod xdrip to run on the pi/Edison then have an all in one solution.
@tim2000s sounds like a great idea. I am just so untrained in programming that unfortunately this was not an option for me
The best that I'd describe my attempts are "hacking about" @C-Ville
Anyone running a Pi0 for portability? Looks to have a low power draw.
how to resolve---am I out of space on micro sd card...
pi@raspberrypi:~/APS401 $ crontab -e
/tmp/crontab.sEpt2Q/crontab: No space left on device
clean your logs
and check with
sudo raspi-config if your image is exoanded to whole sd card
@tim2000s I used to run a pi zero, the edison is significantly better for size and power draw though and it is easier to interface a battery with in a sleek manner
Man, when I went to bed last night, I felt like I was floating. Pretty crazy.
Had a good sleep though!
Morning funny: http://m.xkcd.com/1742/
Heh
"Anything that requires enough configuration that Scott eventually gets tired of doing it manually and adds it to oref0-setup" ;-)
@sjolundjohn re azure limits you described in other channel, did it correlate to using the new setup scripts for first time? see last night convo above, that may be it
@danamlewis yes exactly
It wasn't exactly correlated - it seemed to be more closely aligned with the database upgrade that Azure did as I had the script running 1-2 weeks before
K. Short answer is switch to heroku probably though, long answer is help make the data usage more efficient :)
I would actually put my money on it being related to the Azure upload rather than script to be honest. Since last weekend I hit the quota basically everyday.
I have not looked into Heroku - am I right in understanding that I can replace Azure with Heroku and then get a free tier?
OK fair - that is enough - will make the change tomorrow
I just got the last Nightscout Dev branch, is that recommended still?
$7/month I think but no one is running over data caps even w extreme usage
Yea I believe everyone is still using NS dev
That's the reason because I want to get rid of azure and I would like to run the "new" loop, that's great
@danamlewis thanks
:+1:
@danamlewis 7$ per month what is the limit regarding transfer volume :question:
Not sure off the top of my not yet awake brain
Does anybody have a link for setting up ns in Heroku?
I checked out the stats on Azure. Essentially the issue is that if you run the new script with updates every minute you max out the CPU time you get on the Free level, which is a grand total of 3 minutes. With it set to upload to NS and pull data every minute, it maxed out after about 17 hours. With a five minute pump-loop it was taking 19.
I've set up the $7 heroku and don't have any capacity issues. The stats suggest I'm hitting about 56% of max memory and occasionally peaking at 80% of load.
@sjolundjohn thank you :+1:
I just finished setup, took 10 minutes, very easy.
Will take another 30 to test and configure likely
The deploy to Heroku button is right at the top Readme.md section of git, took me a moment to find it
Wow good to know its that fast @sjolundjohn
Sounds great....I think will try it
It is totally blank without any data, so let's not be sure until working. Off to get my twins so will have to figure out the rest this evening
If you are outside the US and pulling data from share you need to add the
BRIDGE_SERVER EU setting. Works the same as in azure settings but is missing from the steps and I had completely forgotten about it so it took me a little while to figure it out.
That got me too.
I host my NS and mongodb on Google Cloud. Deployment is completely DIY, but with looping it's still only about $5/mo.
Digital ocean is another popular DIY option.
is there like a safe mode boot up for Rp3....intereuppted power seems to have corrupted sd card and it wont boot
it'd be awesome to see some instructions on how to use digital ocean or google cloud instead...I just switched to 55 azure plan as the free plan was exhausted.....but sounds like there ar ebetter options out there
IMO if you need directions for how to self-host, you should be using Heroku.
Looking forward to meeting everyone in SF! @danamlewis @jasoncalabrese @moomoobloo Any suggestions on a nearby hotel for my mom and I to stay?
Administering your own cloud server with a full Linux stack (particularly one controlling a medical device dosing life-critical medication) is not something to take on lightly if you don't already have experience doing it securely.
@moomoobloo I'd love a quick tour of Google. :)
I just tried to deploy a new heroku site....ummm how to I change the ENABLE variables--can't seem to find them...azure might have been easier for me actually
I think I just entered my incorrect bridge credentials
@garykidd Just under settings, reveal config vars
I got mio infusion sets through Solara Medical on Friday. I just opened the shipment. 2 out of 3 of the mio boxes seals were broken, but the lot numbers on the outside of the boxes match the lot numbers of the infusion sets inside the box.
Should I be concerned?
They said that they sometimes "break the seals on infusion set boxes to check inventory"...does this sound like BS?
I switched from Medtronic as a distributor because they treat Medicare patients pretty crappy...hence solara medical...and they actually let me order online for Medicare orders...
Call Medtronic and ask them
Yeah, that's what I was thinking of doing...
@diabeticgonewild regardless, like they couldn't tell how many were in a sealed box. I'll bet they are fine but that does sound like a bs answer.
Medtronic said it was fine...
M
So if I can't boot my pi no chance of recovering as catd...,is that right?
Catd = sd card/rig.
Lots of ways to recover, depending on what other gear you have. SD card reader on another Linux machine? Mount it there and make sure filesystem is OK. Spare SD card? Get raspbian going again and copy relevant files. Etc. etc.
Plus serial console during boot sequence, dropping into U-Boot or whatever it uses.
when we were using Pi's a lot, I found that if an SD card won't boot, it's best to reimage it from a backup or a fresh Raspbian install. having an already-imaged backup SD card really helped too
@scottleibrand : should i use dev or oref0-setup for pullrequests, see openaps/oref0#239
for something small it doesn't matter. but would be better to start with a branch that doesn't have the openaps/oref0#232 stuff in it
you could start a new branch off oref0-setup or dev, and then
git cherry-pick a7262458b2f7721325283fbeb230febc5e57af39
(to get just the https://github.com/openaps/oref0/pull/239/commits/a7262458b2f7721325283fbeb230febc5e57af39 change)
i'll try next time
if you don't get to it before me, I can cherry-pick it over myself
@ecc1 maybe I can try and load sd card on my imac. I suspect the file system may be shot, however. IT seems to just hanf with a blinking "_"
Is U_boot a feature of raspi Jessie?
less than 24 hours looping my azure contingent has been exceeded, web app is stopped, really annoying. The new loop is very data consuming.
@ceben80 Go and make a Heroku app. Get a new email address for it if neccessary
But it is not running 24 h without a break....the free one
No, but Heroku (the paid one) is $7 a month. The alternative is to modify the
ns device to be less data intensive.
That works with autosens?
Yes - I've been running it for hte last two days
what in the ns device makes the new loop so data intensive?
well, there's the fact that oref0-setup.sh is still using dumb
curls for upload, and not using the ns device at all except for BG download
As many megabytes of data will be transmitted. I am very amazed compared to my old loop
But I love the new one except this azure stuff
just switched my nightscout instance to heroku - i find it much easier to configure and understand - and $7/month is obviously very straightforward
:+1:
Now I have attach my Dexcom, N8 to all :wink:
Finally my day ends with a 5.5 A1C :star2:
Hey all! Been scrolling up but didn't see any mention of Animas warning its customers one (or some?) of their pumps can be hacked... has the openaps community anything to do with that? :)
I think most of us saw it. I'm sure Jay was aware of @mariusae's work, but didn't cite it explicitly AFAIK
hmmmm....doubting my share u/n p/w.
where can I login to confirm again?
been awhile since I've done it manually
can someone give me an example of how the variables can be passed to the oref0-setup.sh when you run the .sh file. I was able to get it to work with Scotts old script, but the new stuff is throwing me. I've been testing a lot of times, and it would be nice to do it all at once without the interactive script.
Most of them are printed in the first line.
ah ok
Is that line just invisible? Or is it not detailed enough? Or ?
Ideas/PRs for clarification welcome.
I'll play with what I'm seeing and see what works
it took me a few times to understand your older stuff. So I have someplace to start
what I had to do with my problem last night was do the npm run global-install after I pulled it. Would that have something to do with switching branches?
yeah, when a new branch has a new tool, you have to run that to install it
Hi, I am struggling to get my openaps information into Nightscout. Nightscout is pulling information from a G5 bridge but I can't get the upload working from my Pi. Not too sure where to get help...can somebody point me in the right direction?
have you tried oref0-setup yet?
This message was deleted
Sorry I missed the converstation last night around services,) I'll throw it out there again, the Nightscout Foundation has essentially free machines on Azure. If you're paying for any of it, we'd be happy to hook you up with a VM/Webapp/whatever to make your looping less costly.
@jwedding do we have any way to do that at scale for anyone who wants to use azure with openaps?
@jwedding I’d actually love that. I’ve been fighting with getting Nightscout to run on a Digital Ocean VPS and just about have given up
@scottleibrand looks like you forgot to add the -g to the sort in the oref0-set-system-clock.sh in branch oref0-setup
@scottleibrand I'm not actually using heroku in anger as i'm under the free quota Azure even with Loop/openaps info. But it appears to me that you could use heroku free if you have verified a credit card (extra 450hours free dyno time ) and use uptime robot to keep the site awake. Is there something I'm missing?
maybe. but if you're relying on NS to get BG data to your loop, I would rather pay the $7/mo vs. relying on an uptime robot to circumvent heroku's design
True, it is probably worth it for the peace of mind
so how should I handle?
pi@raspberrypi:~/APS401 $ crontab -e
/tmp/crontab.Ys7Wg8: No space left on device
Creation of temporary crontab file failed - aborting
This message was deleted
This message was deleted
oh wait, this is a Pi
that won't do any good
have you run
sudo raspi-config and expanded the file system?
I believe I did that through the GUI on RP3...but let me that again.
if you've already done, that, take a look at the output of
df -h and cd / && sudo du -xh -d 3 | egrep "[0-9]G|[0-9][0-9][0-9]M" to see what's using up all the space on the disk that's full
368M ./var/cache/apt
377M ./var/cache
101M ./var/lib
8.0K ./var/tmp/systemd-private-f#####################e0ef-rtkit-daemon.service-iujA9G
8.0K ./var/tmp/systemd-private-###################303-rtkit-daemon.service-aJ7GFI
104M ./var/log
685M ./var
102M ./opt/sonic-pi
651M ./opt/Wolfram/WolframEngine
651M ./opt/Wolfram
797M ./opt
101M ./lib/modules
148M ./lib
2.4G ./home/pi/APS401
173M ./home/pi/src
2.7G ./home/pi
2.7G ./home
456M ./usr/lib/arm-linux-gnueabihf
206M ./usr/lib/libreoffice
166M ./usr/lib/jvm
1.3G ./usr/lib
207M ./usr/share/locale
104M ./usr/share/doc
871M ./usr/share
115M ./usr/include/boost
145M ./usr/include
2.4G ./usr
6.7G .
2.4G ./home/pi/APS401
that is excessive
also looks like you did a full GUI install with lots of packages you don't need, like libreoffice, Wolfram, and sonic-pi
dpkg-query -Wf '${Installed-Size}\t${Package}\n' | sort -n will show your installed packages in size order, largest package last (http://unix.stackexchange.com/a/107039)
then you can do
sudo apt-get purge <packagename> on stuff you don't need
what about in my instance directory...how do I drill down in that and remove excess baggage
it'll be the .git directory most likely. cd into that directory and run the
sudo du part there
you can just
rm -rf .git and then oref0-reset-git to re-create an empty one
unless you need git history for forensics on something that happened recently, that shouldn't affect your loop at all
git direcotry under APS401?
ya
not there
it's
.git not git
it won't show up on standard
ls: you have to do ls -la
OK, do you have to pass a qualifier to make it show maybe?
gat it
got it
so after sudo du ....lot of lines withone jumping out...
2359388 ./objects/pack
once you're messing around inside .git it's safest to delete the whole thing
whats the -rf part?
recursive
?
-r is recursive. -f is force
got it
oref0-reset-git from APS instance main directory?
pi@raspberrypi:~/APS401 $ oref0-reset-git
Attempting to fix git corruption. Please wait 15s.
/usr/local/bin/oref0-fix-git-corruption: line 7: cannot create temp file for here-document: No space left on device
Git repo does not appear to be corrupt.
/usr/local/bin/oref0-reset-git: line 52: 4930 Terminated ( sleep 15; killall oref0-fix-git-corruption )
you need to
rm -rf .git first
OK...maybe I did that in the wrong directory...
ahhh
finally
also I
sudo apt-get purge libreoffice-common
seemed to hang on
Processing triggers for gnome-icon-theme (3.12.0-1) ...
but after like 3 mins...it finally completed successfully.
presuming
sudo apt-get purge wolfram-engine will be fine as well?
I guess I can always reload it.
yep and yep
and sonic, and probably some others
that should give me quite a bit more room I would think
gonna make an image copy of card here in a bit so as not to lose all the work (especially with mmtune/ mmeowlink and the like).
I have a couple of SOR modules en route--wonder if they'll behave the same way (I suspect they will). But I will know for sure soon.
pi@raspberrypi:~/APS401 $ cd / && sudo du -xh -d 3 | egrep "[0-9]G|[0-9][0-9][0-9]M"
368M ./var/cache/apt
377M ./var/cache
8.0K ./var/tmp/systemd-private--rtkit-daemon.service-iujA9G
8.0K ./var/tmp/systemd-private--rtkit-daemon.service-aJ7GFI
104M ./var/log
682M ./var
101M ./lib/modules
148M ./lib
173M ./home/pi/src
313M ./home/pi
313M ./home
456M ./usr/lib/arm-linux-gnueabihf
166M ./usr/lib/jvm
1.1G ./usr/lib
207M ./usr/share/locale
104M ./usr/share/doc
834M ./usr/share
115M ./usr/include/boost
145M ./usr/include
2.2G ./usr
3.4G .
:+1:
that's 3.4 Gb used..right
ya.
df -h will show the free
feel free to paste from
dpkg-query -Wf '${Installed-Size}\t${Package}\n' | sort -n if you want more input on things to remove
from 10,000 down...or up depending on how you look at it....
10000 greenfoot
10299 libqtgui4
10440 systemd
10715 gnome-icon-theme
10720 fonts-freefont-ttf
10774 perl
10970 libstdc++-4.9-dev
11332 libruby1.9.1
11680 omxplayer
11853 poppler-data
12000 libgl1-mesa-dri
12033 raspberrypi-bootloader
12810 ri1.9.1
12871 libruby2.1
12955 coreutils
13218 desktop-base
13344 cpp-4.9
13431 pypy-lib
13515 libavcodec56
14087 adwaita-icon-theme
14352 gcc-4.9
15108 perl-modules
16136 iso-codes
16266 locales
17009 libc6-dbg
17942 samba-libs
18542 git
18681 binutils
19400 gputils-common
19671 libflite1
20709 libgtk2.0-common
21367 g++-4.9
22684 libgtk-3-common
24828 libpython2.7-dev
26095 libwebkitgtk-3.0-0
26531 libicu52
27472 libqtwebkit4
32846 libllvm3.7
32888 pypy
33041 libraspberrypi-doc
33224 freepats
34217 libicu-dev
44964 nuscratch
45673 scratch
47401 libpython3.4-dev
91230 libboost1.55-dev
104882 raspberrypi-kernel
168481 oracle-java8-jdk