Absent IP weather station prevents all logging?

[skyewright]

It appears that the absence of an IP based connection (i.e. impossible to connect to that IP) to a weather station (or at least specifically to a VP2) prevents logging from all stations on the Meteohub.

Scenario:

I have 2 Meteohub, Mb and Mc.

Mb has 3 stations defined:
0) USB RFXCOM
1) Plug-in via TCP/IP on Mc
2) VP2 via TCP/IP (the 'far end' of the IP is VirtualVP running on a PC)

Mb) has 6 stations defined:
0) TCP/IP RFXCOM
1) System station
2) WMR200 (stopped, and not actually plugged in)
3) Custom Plug-in
4) Custom Plug-in
5) VP2 via Meteohub TCP/IP (i.e. the other meteohub)

I have found that if the TCP/IP connection for Mb2 is unavailable (e.g. that PC is rebooting) ALL logging on Mb ceases.
I tried ticking "stop" on the Mb2 station then saving, but still no logging at all (meteohub log showed the other 2 stations as having started okay).
I tried manually stopping and restarting logging - still no logging at all (Mb2 still had 'stopped' ticked).
I tried rebooting - still no logging at all (Mb2 still had 'stopped' ticked).
I tried changing the connection from TCP/IP to Meteohub TCP/IP (because all the while the Mc5 station was okay, except of course that it wasn't receiving data) - still no logging at all (Mb2 still had 'stopped' ticked).
Finally I changed the connection type of Mb2 to serial (on which there is nothing) - with that Mb0 and Mb1 data started flowing again. Phew!

The PC to which the VP2 is actually connected and which is running VirtualVP is now available again (the root of the problem was that the latest set of WinXp updates took over half an hour to install on the low energy PC!). Mb2 is now set up as TCP/IP again, and all data is flowing. :)

So, I know what to do in future if there is going to be a "planned outage" (i.e. change Mb2 to serial beforehand), but what if there is a problem when the Meteohubs are unattended?

It would be very nice if Meteohub could handle the situation a little more elegantly. :)

Boris: If you would like access to my Meteohub(s) you are welcome. I could arrange an 'outage' for you to observe if it would help.

[skyewright]

It would be very nice if Meteohub could handle the situation a little more elegantly. :)

I'm delighted to report that Meteohub 4.5g handles this much better. :-)
Presumably as a result of the change described as:

Improves time-out handling of weather stations connected via TCP/IP.

Thank you. :cheer:

[HeinrichH]

My Meteohub has also connected 2 weather stations by IP.
One station is the RFXcom receiver and the other the Davis Vantage IP logger. The Vantage logger has problems now for about 4 weeks and MH is not able to connect, sensors report "never" but the Oregon sensors connected to the RFXcom are logging quit well, the only problem is that once a day the data storage is filled for 100% and the data send to WU is stopping, normaly it's about 32% used. I'm running version 4.5f on the moment and the PC is running Vista Business, it's a low power Acer PC

[skyewright]

I'm delighted to report that Meteohub 4.5g handles this much better. :-)

...but perhaps not perfectly... :(

Earlier today I made a calibration adjustment on the VP2 console. Half an hour later I noticed that the Meteohub that reads from the console (via VirtualVP) had not seen any VP data for 30 minutes and was very busy (System Load up around 7 instead of the usual value of less than 1).

Looking in meteohub log showed a whole load of entries like the one below from around the time of the console adjustment:

logger (02.11.2009 10:32:48): child process received termination signal (11).

then

logger (02.11.2009 10:32:57): child process received timeout signal (14).
logger (02.11.2009 10:32:57): data logger stopped.
logger (02.11.2009 10:32:57): data logger started.
logger (02.11.2009 10:32:59): connect station 0 (RFXCOM via USB serial).
logger (02.11.2009 10:32:59): connect station 1 (Plug-in via TCP/IP Meteohub).
logger (02.11.2009 10:32:59): connect station 2 (Vantage via TCP/IP).
logger (02.11.2009 10:32:59): unexpected 0 bytes delivered from weather station 1 (Plug-in)
logger (02.11.2009 10:32:59): disconnect station 1 (Plug-in).
logger (02.11.2009 10:32:59): station 2 (Vantage), syncing.
logger (02.11.2009 10:33:29): connect station 1 (Plug-in via TCP/IP Meteohub).

It's hard to say how many "child process received termination signal (11)" lines there were as there were at least enough to completely fill meteohub log window aside fom those last few lines.

What still hadn't appeared in the log around half an hour after the 02.11.2009 10:32:59 line was anything like the usual lines about rain collector size, etc. and a "station 2 (Vantage), ready logging data" entry, nor had Meteohub aborted the attempt to sync, then tried again.

I needed to reboot to get back to a good connection...

PS. My assumption is that this issue and the one reported by HeinrichH are entirely different matters.

[HeinrichH]

For me the problem on the moment is that I'm working in Saudi Arabia, living in the Netherlands and the station is based at my second home in Portugal....So it's difficult to do any hardware checks and/or repairs now but next week I'm in Portugal for a couple of weeks so I can repair the Davis station.
Since the abcense of the Davis logger I have to reboot Meteohub twice a day but I don't want to disconnect the station because a lot of my graph defenitions are using the Davis sensors