CumulusMX "Data input appears to have stopped"

[water01]

3026 is the latest CMX build.

Edit : Cross posted

[madsandrup]

Hi Steve,

I know this thread is about RPI, but I have now got this same error twice since updating to 3026 on my Windows 7.
Both times I have just restarted CMX and everything works again.

[steve]

Yes, I'm afraid I introduced a bug when I added some code to handle the receiving of bad data. I should have a fixed version out soon.

[madsandrup]

I have now installed 3028 and will keep an eye on this build.

[n1iwv]

Data stopped last night around 10: P.M.
I can tell by looking at the "trend" graph and it flat lines.
I've attached the latest MXdiag file.
Build 3028

Thanks
Ed

[steve]

Some kind of communication problem. It stopped receiving data, and shortly afterwards, when it tried to use the TCP connection to the logger, it was told that the connection had been closed. You had ftp communication errors at the same time, so this was presumably a more general communications issue. You seem to be having regular communication issues with your system. It's become clear that WLIPs are a complete pain (although mine has been running for 11 days now without a glitch) - I'll see if I can add some more retry code to attempt reconnect at intervals (rather than just retry the reading of the data) when the system is having communication issues.

[madsandrup]

I haven't had any problems with this issue since updating to 3028.

[stewartwlewis]

Hi
I have been running MX for a while now with no issues but over the last week it started just stopping. Closedown and restart and everything is fine then it stops again. Seems to be the same error as this post.

I am running RPi with MX3030 and the included version of Mono as using standard English settings.

The console shows
_wapi_connect: error looking up socket handle 0xf

Though earlier crashes dint show anything on the console at all.

I have attached the diag logs. Some connection errors on other days but fine before crash.
This seems to have happened at 2015-11-11 14:07:10.296 when the Twitter task was cancelled.
This agrees with the trends graph that stopped about then.
Trying to access the local website on the Pi and the browser just hangs with no output.

Any help would be great thanks.

[steve]

It was still reading data right up to the end of the file (at 18:17) and updating WOW with updated data, so that part of it was apparently working fine, and hadn't stopped. There's no 'data input appears to have stopped' message anywhere in the file, so it's not clear to me what you are referring to when you say "seems to be the same error as this post". This thread was about problems reading data from an IP logger, which I think I have now fixed (I have an IP logger and a Pi, and MX runs flawlessly on it).

Can you clarify exactly what it is that you see? Was it still writing data to the log files up to 18:17, for example? Perhaps the problem was that the built-in web server stopped receiving connections?

[stewartwlewis]

Hi Steve
Yes it was still posting and is still posting but the webserver is not responding and the web site has not ben updated. The last update was 14:16. The log from 16:17 just repeats the same loop 2 command and is doing the same now.

Thanks for looking at this, is there any more information I can post that might help?

[steve]

It looks like it was doing very little other than reading the data. For some reason the system timer which should fire periodically and cause all of the other tasks to take place wasn't firing. Unfortunately I have no way of knowing why that was the case, and there's no other information from Cumulus that would help. Your system logs may show some indication of a problem.

[stewartwlewis]

OK thanks Steve.
Will have a look but the Pi seems to be working just fine. Noticed mistake in last post. Log is doing the Loop 2 loop from 14:17.
Not material but posted here as the error on the console is exactly the same as the one at the start of this thread.