A Share Strategy During the Mlab Update
An unexpected problem with my Nightscout site leads me to put an added safety net in place before this week’s maintenance window begins.
When I used to teach college-level English 101, “red herring” arguments and “straw man” arguments were always problems that came up with student essays.
These concepts come to mind today for some reason as I watch the CGM in the Cloud community prepare for the Mlab 3.2 update. Community members are obviously concerned about losing their data, about sites failing, and about their T1Ds having a problem while remote monitoring is unavailable. And as community members try and do necessary updates, they wonder if every problem they see today is related to the pending update.
At this point, the updates have not started. A problem with your site today is simply that… a problem with your site today.
Last year’s Mongo 3 update was infinitely more complicated for people. Hopefully we learned a lot from the experience. We all are hoping this year’s update goes much more quickly (on the Mlab side) and much more smoothly for Nightscout families. Until the update happens, however, we can’t really be sure.
Unfortunately, for some families, touching their sites or setups leads to problems. The problems can not be explained, really, by the fact that an update will happen tomorrow (in the future). But problems do happen. And right now, with the looming Mlab update in their minds, issues of cause and effect get a bit cloudy.
A Problem and a Solution
As I left the house this morning, scrambling to get my late middle schooler out the door, I realized that my Pebble wasn’t working. His was because he is using the nice new xDrip offline face (for xDrip beta 6 and beyond). He had a number, so it was really last-minute when I realized I didn’t have a number. (I have had countless Pebble issues due to recent updates. My watchfaces say they are not connected or have no bluetooth more often than they show me a number.) Reloading the face didn’t solve it. Closing and restarting my Pebble app didn’t solve it. Checking my Nightscout app (iOS) showed me, alarmingly, that it had been 4 hours since it had updated. (I had been up and working during those four hours, and I hadn’t noticed that our data was not uploading. It had worked all night.)
Having seen a few rogue issues this morning in the CGM in the Cloud Facebook group, I wasn’t sure what was up, but something was. As we rushed out the door, I pulled the uploader phone back out of the bag and checked the error log. Sure enough, there were clear rest api upload errors indicating data wasn’t making it to Azure.
My student was 74 mg/dL at the time. We have no nurse. Our plan is not in place for this year. In reality, there is no oversight at school and no awareness of my T1D walking the halls as a T1D. My ability to monitor remotely makes these realities somewhat tenable on a day-to-day basis.
Seeing my site down, in advance of the Mlab update, however, was a concern.
Thinking through the options is always important.
What did we do? As I drove the eight minutes to school, I talked my student through downloading and installing the Dexcom Follow app on my phone. At stoplights and stop signs, I configured his xDrip app to also upload to Share servers. When the Follow app loaded and indicated that the only way to configure a follow account is through a new email invitation, I resent the Follow invitation to my primary email.
It didn’t show up in email as immediately as we would have liked. So I sent another invitation to a secondary email address.
Uncertain that I had our login info correct, and worried we were going to run out of time before I could confirm that the upload was working, I had him use the browser on my phone, Google “Dexcom login,” and test the login information I had entered. (It worked.) Then I worried that the serial number in the upload to Dexcom Share section might not be correct since we did recently change receivers. I walked him through checking that. Our login was fine. Our serial number was correct.
Once the Follow invite arrived in email, he was able to click through and establish my Dexcom Follow.
With a sigh of relief, I saw the current reading come through as I turned down the street to school.
He was on time.
And I had oversight.
Once home again, I checked Mlab status and Azure status. Neither showed outages that explained what was going on with my site. Initially, a restart in Azure seemed to not solve the problem. That was an even bigger concern. But, after sufficient wait time, the site loaded up again, and we were back in business.
But… Dexcom Follow is also along for the ride this week.
I hope that when my db goes down and is updated this week by Mlab that the downtime really is in the neighborhood of 5-10 minutes and that a simple restart restores my site afterwards. But… we have a G4 Share… and for the moment, taking advantage of what that offers us makes sense.
If it was me… and I had a G4 Share and my T1D carries an Android with xDrip, I would do three things:
- Configure xDrip to upload to Share. (You can always turn if off after this week’s maintenance window.) Be sure to send yourself the invite from xDrip, install the Follow app, and confirm things work in your Follow app. (If you, personally, use Android, you can also Follow using Nightwatch, no invitation required.)
- Configure the xDrip offline face for the T1D’s Pebble. Beta 6 works with a new xDrip face. Earlier betas can use the xDrip face that is available in the Pebble store. With this face active for the next four days, I know that even if I lose sight of his data, my T1D will have numbers on his wrist.
- Set up one of the other watchfaces on your own Pebble to show Share data. This way you can easily switch to a new face if your site is down, and tap into the Dexcom Share data.
A Lesson in Preparation
The Mlab updates that will begin tomorrow didn’t cause my problems today. That my coffee maker also conked out today also wasn’t really caused by the coming Mlab update, although I am really going to be unhappy tomorrow morning when the Mlab update starts hitting and I haven’t had sufficient caffeine. Sometimes things just happen in strange sequence, and it is easy to leap to conclusions.
The outage this morning made me stop and take a step that will give us peace of mind this week.
I am sorry that this step is not available to all Nightscout families — because G4 with Share is not available everywhere. And I hesitate to suggest these changes in the group because, in truth, Kate and I, alone, can’t help every family walk through the steps to turn Share on. Taking these steps really means you need to be willing to use the general information provided (in my story) and the feature available (in xDrip) and put some additional settings in place to create an extra layer of safety net.
While I can’t walk every family through this, I know these suggestions may help someone. So I am putting them here. They might help someone take advantage of the tools available to help make a short maintenance period easier.