Cumulus MX and Meteobridge

[mcrossley]

Guy, does WeatherLink download archive data? Or is it just running 'realtime'?

[steve]

Yes, I think Boris is the only one who can get us any further. He hasn't been on the forum for a long time - I actually have a message sitting in my outbox for him from last March. I've added him to the beta test group in case he does turn up.

[grwkak]

Steve - Let me contact him through the Meteobridge forums - if thats OK with you.

[grwkak]

Mark - I am not sure what the software really did as I don't really use it. I ran it simply to check if I could see any data (did I have a communication issue that was not software specific)- and after 5 or so minutes (the logger interval) some showed up.

Guy

[grwkak]

Posted in the Meteobridge forum, described the issue and showed the diags output.

Response below:

"When the PC program takes over, the console might still be in an mode triggered by Meteobridge to report some type of data (LOOP packets etc.). So Cumulus cannot not expect that it will find an empty receive queue where only data it has requested will be there. After a few more couple of seconds, no more Meteobridge requested data will come in, as Meteobridge stops requesting from the console when the PC program has made a connection.

So consuming data not requested before starting its own protocol and doing some initial looping on false tries might be a good idea for Cumulus and this is not just something that can happen in context of Meteobridge. When another PC program (or previous call of Cumulus) has requested 100 LOOP packets and terminates then. The PC program started next also has to handle this situation.

Will Cumulus MX make use of the LOOP2 packets as well or is it still stuck to the antique LOOP protocol only (makes a difference in pressure readings etc)?"

I know when using Cumulus it took a few seconds to establish connection and see data.

Guy

[Adrian Hudson]

I used to use Metohub (not Meteobridge) with "another weather program". Meteohub is like Meteobridge except it keeps a record of the data itself and can generate graphs/reports etc. Indeed it is written by the same company and obviously uses much of the same code. it got me out of many a tight spot when the other weather program had failed. I could regenerate the data for the other weather program from meteohub using scripts I made. I even made a failover web site that kicked in if the other weather program had failed - once again.

I stopped using Meteohub when I saw the light and came to using Cumulus - I wish I hadn't really - not that I need it to make up for any deficiencies in Cumulus - far from it. But it was a nice belt and braces solution.

If you do happen to get this working I may very well play with Meteohub once more.

[steve]

If you do happen to get this working I may very well play with Meteohub once more.

This is specific to the new 'pass through' facility in Meteobridge. If I add the code necessary to work around the issue with Meteobridge, it doesn't mean that Cumulus will work with Meteohub, that would effectively require support for a new station type.

The problem appears to be specific to the way the connection is handed over untidily. This does not happen with direct connections. I can close down Cumulus MX with my serial logger and start it up again immediately without hitting this issue. The only issue that can happen is where the console sends the entire contents of the logger, but it doesn't stop it working at all as with Meteobridge.

I can do the same with a TCP/IP connection to Virtual VP.

It would be useful to know whether this happens with a direct connection to an IP logger. That is - can you reliably and repeatably stop MX and start it again immediately?

[Adrian Hudson]

If I add the code necessary to work around the issue with Meteobridge, it doesn't mean that Cumulus will work with Meteohub, that would effectively require support for a new station type.

Ahh, okay, once again my misunderstanding. {bows out of conversation...}

[grwkak]

Latest build 3017 seems to fix these issues - thanks!!!

Guy

[grwkak]

May have spoken too soon - data collection starts but freezes. Downloaded 2018 to try.

Guy

[grwkak]

Looks like 3018 has the same issue. Displays data at first but data stays unchanged. Still - no zeros for data so progress...

Guy

[steve]

It's still getting garbage (leftover loop data) from the untidy handover of the TCP/IP connection. And I've even added code to stop the sending of loop data, and to read off any old data for a short while at startup. Perhaps it needs to do it for longer. I would have to make it an option, though, as this device is the only one that needs this, and it would delay the startup for everyone.

[steve]

I forgot to mention that I have now made the initial wait time configurable. Add a line to the [Station] section of Cumulus.ini:

DavisInitWaitTime=5000

The value is in microseconds, so 5000 allows 5 seconds for any 'old' data to be received, which should be long enough for a loop packet. A smaller value might work (200 ms is the default).

[grwkak]

Thanks - will give that a try.

Guy

[grwkak]

Still having issues - appears to connect and collect data at first and then does not seem to update. Let me play with the delay settings a bit.

Guy