Server restarts this PM?

Has there been a burst of server restarts this afternoon? My agents appear to be restarting every 5 mins or so for about the past 45 mins. (I’ve got some notifications tied to server restarts, which also notifies that me that my devices are online, and have just been flooded over this period).

We were experiencing some backend issues earlier this afternoon - but they appear to be resolved now.

We are continuing to monitor the situation, and will post updates if required.

just checked my project - it is not responding to the ajax calls - so http.onrequest and maybe more is not working…

Just as a reminder (or maybe you’re hearing about this for the first time) - we have a twitter feed for system status (and we’re getting better at updating it as we experience issues).

Thanks @bi … it just got unusually funky rather quickly. I’ll just keep the notifications temporarily disabled in my agents 'til things stabilize a bit. Thanks again for the update.

Oh man … I was hoping I’d never have to sign up for Twitter … (showing my age again I guess :slight_smile: )

You don’t have to signup for Twitter - you can just go to the provided link when things look weird to see if we’ve posted about any problems :slight_smile:

Agents should be working again (though we’re still investigating the root cause of the issue) - if you’re still experiencing problems please let us know.

I am having issues too just noticed by Xively feed has stopped and my Agent is no longer responding requests. Reboot has not fixed the issue. My code is still logging so its working at that level

Definitely still having issues, I can now send requests to my imp and they are being received but still nothing being sent to xively from my Imp (using the blueprint still for this). Anyone else having this issue?

I am also seeing the same issue for the past few hours. The Imp IDE logs data properly, but Xively is not receiving data

me too!!! help please xively is not working… :frowning:

Same issue here. Xively not working.

in my web app this shows up as a “is not allowed by Access-Control-Allow-Origin” error so possibly a configuration has changed.

…edit… nevermind. I had the wrong - actually blank - URL in my request so I was not even sending my request to the correct place.

so for my stuff the agents are working.

See Cosm post - Xively is still not working as they are blocking requests from our outbound IP. Likely we crossed some threshold and their automated systems kicked in. We’ve emailed to get them to whitelist us.

Doh! It’s an Imp DoS attack! :slight_smile:

Precisely the issue I have with cloud computing, too many dependent systems make for an unreliable and flakey solution. Just as well I’m just doing play stuff, if this were a production solution with real customers my business would be taking a hit here.

@ChrisRusty - we actually have a separate group of servers for our customers in production (for exactly this reason).

Our developer community seems hell bent on testing and pushing our services to the limit (which is great, by the way) - but leads to a somewhat less stable environment.

By the time a customer gets to production, their code has typically gone through enough revisions and testing that it’s fairly stable, and not doing potentially dangerous things (like making hundreds of outbound web async web requests a second). As a result - the production servers are more stable.

I don’t know how you would build a good Imp-like solution without a cloud element… that just the way things are/will be done.

As I was thinking about Imps and DoS attacks, I actually came to the conclusion that because Imp’s use a cloud service, they would actually be much less likely to be targeted for a DoS attack than a device that was directly controlled.

I find Agents, the web IDE, and Electric Imps’ ability to easily pass down firmware updates as extraordinarily useful. I do more and more things with agent programming everyday, and am eager to see what new features are added. No service is going to be without glitches

Now, certainly, the more services you use, the more risk you take. For that and other reasons, I am working on trying to do everything I want to do within the Imp and the agent for each project… even things like charting data, that Xively does very well. IMHO, the cloud connected benefits of the Imp far outweigh any risks or negatives.

Don’t get me wrong think the Electric Imp cloud approach is great. It is cloud to cloud dependent services I’m talking about here. There was a small outage about 24 hours ago. Thousands of wee imps got restarted and a good many of those sent requests on restart to Xively which in turn detected this as a DOS attack. Least that’s what I think happened. I lost my Solar heating monitoring and control app for an entire day as a result. I agree the more Xively type functionality you can built yourself the better. I’m just using them for the charts and triggers as a stop gap till I built my own hosted solution.