CumulusMX lost connection, does not reconnect until restart

[jmelan]

Hi, I am new to cumulus, thanks for the great program!

I am using build 3036 on Ubuntu 14.04 LTS connected via passthrough to a meteobridge attached to a Davis USB logger (Vantage Pro 2 Plus). Everything was working fine for a week or so.

Yesterday the connection to the Davis logger appeared to have been briefly dropped by meteobridge and reconnected. I have meteobridge uploading to WU and there are no dropouts there.

Code: Select all

Meteobridge log
logger (30.01.2016 17:41:26): child process received timeout signal (14).
logger (30.01.2016 17:41:26): disconnect station 0 (Vantage): exit
logger (30.01.2016 17:41:26): data logger stopped.
logger (30.01.2016 17:41:26): Historical data loaded (24 records).
logger (30.01.2016 17:41:26): data logger (version 5.0m, build 8944) started.
logger (30.01.2016 17:41:27): connect station 0 (Vantage via USB serial).
logger (30.01.2016 17:41:27): wakeup sent.
logger (30.01.2016 17:41:27): internal logger activated.
logger (30.01.2016 17:41:27): syncing.
logger (30.01.2016 17:41:37): gettime request sent.
logger (30.01.2016 17:41:37): time offset of Vantage: -8 minutes
logger (30.01.2016 17:41:37): eeprom request sent.
logger (30.01.2016 17:41:37): rain collector size 0.01 inch per tick.
logger (30.01.2016 17:41:37): transmitter 1 active as station type (#0) "ISS".
logger (30.01.2016 17:41:37): firmware version request sent.
logger (30.01.2016 17:41:37): firmware version 3.15
logger (30.01.2016 17:41:37): no reading of Vantage's internal logger.

At the same time, the connection to CumulusMX was dropped and an error message was generated (below). However CumulusMX did not reconnect and I had to manually restart the program after which it reconnected without an issue.

Code: Select all

System.IO.IOException: Write failure ---> System.Net.Sockets.SocketException: Connection reset by peer
  at System.Net.Sockets.Socket.Send (System.Byte[] buf, Int32 offset, Int32 size, SocketFlags flags) [0x00000] in <filename unknown>:0 
  at System.Net.Sockets.NetworkStream.Write (System.Byte[] buffer, Int32 offset, Int32 size) [0x00000] in <filename unknown>:0 
  --- End of inner exception stack trace ---
  at System.Net.Sockets.NetworkStream.Write (System.Byte[] buffer, Int32 offset, Int32 size) [0x00000] in <filename unknown>:0 
  at . (System.Net.Sockets.TcpClient , System.String ) [0x00000] in <filename unknown>:0 
  at . () [0x00000] in <filename unknown>:0 
  at System.Threading.Thread.StartInternal () [0x00000] in <filename unknown>:0 
**** An error has occurred - please zip up the MXdiags folder and post it in the forum ****
Press Enter to terminate

Does the program not try to reconnect on its own if an error occurs? Is there a way to set it to automatically reconnect/restart in case of an error? Or is there another workaround for this type of error.

I can post the mxdiags folder if needed

Thanks

[steve]

It currently does not automatically reconnect, for 'standard' setups connected directly to an IP logger this does not seem to be an issue. But it's something I intend to look at for a future build.

[txweather]

@jmelan,

This seems to be a common issue with meteohub and cumulusmx.

I have the exact same issue!

I use Debian 8 and Meteohub.

[steve]

I have the exact same issue!

I thought that the problem you were seeing was that Meteohub just stops responding - in particular when it gets a LOOP2 request? And disabling LOOP2 has worked around the problem for now? Are you seeing the TCP/IP connection being disconnected as well, resulting in that "Connection reset by peer" exception in MX diags? It doesn't really matter, because that clearly is an issue as well, I'm just confused by what you're saying, if I haven't understood your problem after all the time I've spent on it.

I've added code to the next build to retry the LOOP2 and wait longer, but I don't know when I'll get around to handling the TCP/IP connection actually being disconnected.

[txweather]

I have the exact same issue!

I thought that the problem you were seeing was that Meteohub just stops responding - in particular when it gets a LOOP2 request? And disabling LOOP2 has worked around the problem for now? Are you seeing the TCP/IP connection being disconnected as well, resulting in that "Connection reset by peer" exception in MX diags? It doesn't really matter, because that clearly is an issue as well, I'm just confused by what you're saying, if I haven't understood your problem after all the time I've spent on it.

I've added code to the next build to retry the LOOP2 and wait longer, but I don't know when I'll get around to handling the TCP/IP connection actually being disconnected.

Steve,

Sorry. I been so busy at work that I have not been able to come and post here. But yes the problem exist just not as frequent as before. But like the OP here after a while of running catch the same exception as his.

I attached my logs just in case you need them.


Edit: Is too big for an upload so I put it on my web server. Please grab it when you get a chance http://www.wxdev.txweather.org/tmp/MXdiags.tar.gz

[steve]

But like the OP here after a while of running catch the same exception as his.

I can't see a "Connection reset by peer" exception anywhere in those files.

[txweather]

But like the OP here after a while of running catch the same exception as his.

I can't see a "Connection reset by peer" exception anywhere in those files.

Steve,

I can see it on the console though.
I will post it once it happens again.