Are agents down?

Don’t seem to be getting response from any agents?

I think so…

A few people seem to be experiencing this right now…

If you haven’t already, try refreshing your planner/ide browser window - your agent may still be running, but not reporting to the web page properly if it’s been open since we rebooted a server.

If this doesn’t work, let us know - we’re looking into the issue and will keep you up to date as best we can.

not showing in the IDE is not the issue, it doesn’t run code and it doesn’t update new modifications

One other thing that has been suggested - please try power cycling your device… if it disconnected (and didn’t reconnect for some reason) then updating your agent code will have no effect (we require the device to be online for agent code to be updated).

If that doesn’t fix the issue, can you please email or message me the agent URL for your broken agents (we’ll be able to dig into our logs with this information and see what’s going on).

My agent/imp are back online.

ok agents work again, communications as well.
Still lost a lot of code :frowning:

We didn’t show any agent issues in the last day, and our agent tests here have been working fine.

There did appear to be an IDE-related issue at one point - were the issues related to debug of agents or already-running agents not behaving themselves?

Things are working again now.

Just for the record I wasn’t using the planner or updating any code. Devices had been off, switched them on to use them and see expected behaviour but nothing happened. Just tried them both again now and they all work again. Not been into the planner at all today.

There did appear to be an IDE-related issue at one point - were the issues related to debug of agents or already-running agents not behaving themselves?

Both I think but I’m not totally sure. But here’s what I saw if it helps.

Last night my agent seemed to stop responding to incoming requests. The IDE was still alive. I added a server.log to see if it was seeing anything. Hitting Run initially did nothing (and I lost the added code when I refreshed the IDE – would be nice ot have a “Save” button). But later hitting run would generate the “Device Booting…” message but still my agent wasn’t getting incoming packets. It never logged anything and my application never got any responses. This condition persisted until this morning.

In addition power cycling the imp last night resulted in it being stuck in the “connecting to server” loop of Orange, Red, Off.

I saw the same issue but I believe it was planner related. Devices I had edited had non-working agents but devices I hadn’t touched seemed to be working fine.

As a general comment, for end-user products containing imps to be mass deployment ready, they absolutely need to be able to reconnect/reboot/resync/… autonomously in all scenarios after a infrastructure related issue occured . A glitch in the environment once in a while is acceptable (although not nice) and understandable in this life cycle stage of the Imp product, but we will have no way to cycle power or stuff like that since we have no connection if the Imp is down and we can’t call thousands of users to ask them to cycle the power on their device.
Can you please take that as a top priority in terms of field readiness of the product ?


We’re not aware of any major issues with imps failing to reconnect/resync (or connections staying up with the watchdog workaround needed for the current release) - obviously this is paramount for many applications. We’ve seen some wifi issues with the current release, but feedback from these people testing release 23 has shown this issue improved/gone (our internal test units have been connected since we loaded the code months ago despite internet issues and server upgrades).

I personally have devices that have been online for months without being touched.

Do you have examples of devices falling offline?

seems like agents are down the last few minutes? my imps/agents dont work and when i try my agent it keeps trying to connect. Are agents really down or did any of my modifications break something? thanks

There were some sporadic agent HTTP input issues today, yes (agents and output were ok). We’ve got some new developments in that area soon which will help a lot…