ecowitt_http versions: need advice

[Dr__Bob]

Hi All,

I currently have a GW1000, which works fine, except that my raspyberrypi (rpi) periodically loses contact with it.  Of course, when that happens, I lose all the data that the GW1000 would otherwise have sent.  While I try to debug what's going on with the rpi, I figured I might as well pick up a GW3000, since it can have an SD card and allow for catchup when the rpi can't get the live data.  I assume (always dangerous!) that this is a valid reason for trying to "catch up".  It appears that the ecowitt_http driver is ideal for this, since there is the claim that the catch up mode works.  However, it also appears that there are two different forks, which may have diverged significantly.  One is apparently the original, created by Gary and now hosted by Ian Millard.  There doesn't appear to have been any development on that version (or rather, there was, but everything has been rolled back to Gary's original).  The other is maintained by Werner Krenn, and appears to have a *lot* of additional stuff by now, necessitating, I believe, a new database schema.  In addition, at least to my humble abilities, the installation of Werner's version seems involved, with, eg, plugins, and new skins.

(Sorry for the length of this post!)

Bottom line:  does Gary's original version, hosted by Ian work?  Does the catch up functionality work?  Or should I bite the bullet and try to figure out which of the numerous  schemas and database update scripts I should use, and try to get Werner's driver working?

Thanks in advance for any suggestions!

Bob

PS:  I am not a total newb.  I have, in the past, managed to debug a piece of MWall's cmon package, so I know a bit of programming and weewx.  It's just that I got a little overwhelmed by Werner's structure.

[vince]

You're going to have to quantify "loses contact with".   Is the gateway or pi rebooting ?  Is your home wifi all polluted by misconfigured neighbors ?

Gary's GW1000 driver is rock solid for what it does. Rather than worrying software at least initially I'd probably start with working it as a network issue and perhaps catchup isn't critically needed.

Why are you losing connection ?   Is the GW1000 dropping off the network ?  Or is the pi ?  Or is one or the other rebooting ?

Do you have ip addresses set to be reserved in your router for both boxes so that they always have a predictable ip address ?   Do you have the gateway ip address set in weewx.conf ?

What model pi are you running ?  For me any pi zero has always been 'very' unstable trying to stay up over wifi.  I turned all my pi zero off for this reason.  Anything pi3 or later is great.

Lastly - the GW3000 has wired ethernet too as do all pi3/pi4/pi5, so if you can use ethernet for the gateway 'and' the pi, you will likely have a much more stable setup.  Pretty much everybody these days is surrounded by misconfigured neighbors disturbing 2.4GHz wifi and short of out-radiating there's not much you can do there typically if the underlying issue is messed up wifi.

Start by dropping the GW3000 in as a replacement for the GW1000 and see if that helps any.  You might not to do much more than that...

[Graham Eddy]

my experiences with ecowitt gear using wifi interface are all bad - unreliable links that don’t restart properly. the ethernet interfaces, in contrast, have been very stable.

(i am in a noisy radio environment, but that is nearly all just me :-)

⊣GE⊢

[Dr__Bob]

The error that I get is:

ERROR user.gw1000: Failed to obtain response to command 'CMD_GW1000_LIVEDATA' after 3 attempts

Sometimes it recovers immediately (after weewx automatically restarts), but other times it just keeps cycling with the same error message.  During all of this, as far as I can tell, the GW1000 is up, as it also sends data to Wunderground.  And my raspberry pi 5 is up as well, since I don’t see any network errors in the system journal.  Alas, my setup doesn’t really allow for hardwired Ethernet, so I have to rely on wifi.

[vince]

You might try dropping the GW3000 into service then and try to site it close to your access point. I don’t have my gateway posting to anywhere at all, as weewx does that fine for me.

[Rainer Lang]

I think there are a few things to consider
1. is your GW3000 on latest firmware ? 1.1.4 that is
2. one should increase the (probably hard coded) number of retries in the driver from three to 5 or 10 (depending on the time gap in between - but within the default reporting period of 300 seconds
3. a WLAN can be a tricky beast, prone to many possible interfence factors - hence the Ethernet connection is definitely more stable

I had such issues as described recently more often - after the number of WLAN devices had increased significantly - it wasn't the live data but the MAC address request by the driver, but doesn't matter - one of the requests either fails completely or cannot be reestablished in time - then the driver and subsequently weewx exit.
My point 2. above could help here so the driver doesn't give up so fast. This timeout is somewhere in the Python code.

@Ian, @Werner - it might be a good idea to make the retries-when-unsuccessful a parameter to be set in weewx.conf with the default value 3 for the EcowittHttp driver.

--
You received this message because you are subscribed to the Google Groups "weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to weewx-user+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/weewx-user/ff5dc3e5-26c9-4e2d-b965-c59b71d6a770n%40googlegroups.com.

[Ian]

@Rainer, I have acted on your suggestion and now DEFAULT_RETRY_WAIT = 3

Ian

[vince]

Rainer - with Gary's original driver if you set loop_on_init=True and hard-set ip_address=x.x.x.x in the [GW1000] block things are very reliable for me on a GW1200 gateway.

My logs indicate that I have had only one occasion of seeing the Failed to obtain response to command 'CMD_READ_STATION_MAC'  in the last month and that was when the network was actually down when I was upgrading the firmware on my network devices.   Weewx restarted automatically with no problems.

Previously with a GW1000 gateway I did see that issue occasionally but loop_on_init was the solution there as well.

[Dr__Bob]

Hi all,  just to update things.  I did two things (almost, but not quite, simultaneously, which is never a good idea…):  I changed out my “free” Eero 7 router for an Asus RT-BE58U router and I replaced the GW1000 with a GW3000.  Time will tell whether this is more stable or not. The Eero may be a perfectly fine router, but it liked to hide too much under the hood.  I have a better idea of what’s happening with the Asus.  The GW3000 is also closer to the router, which should help. My next step will be to replace the gw1000 driver with Gary’s/Ian’s ecowitt_http driver. I’ve already installed it and did a test run, so I know that it can talk to the GW3000, so maybe over the weekend I’ll swap drivers as well.

Anyway, thanks guys for the help!

Bob