Topic: New GW2000 with flaky wifi

[broadstairs]

I bit the bullet and went for a GW2000 to replace my GW1000. Now wishing i hadn't  the GW2000 is running V2.2.3 firmware.

This GW2000 seems to be flaky on wifi in that I keep seeing this intermittently in the CMX log file

Code: [Select]

2023-05-26 16:02:29.075 DoCommand(CMD_GW1000_LIVEDATA): Error - Unable to read data from the transport connection: Connection reset by peer.
2023-05-26 16:02:29.075 Attempting to reopen the TCP port
2023-05-26 16:02:30.076 Ecowitt Gateway Connect attempt 1
2023-05-26 16:02:30.144 Error opening TCP port: Connection refused
2023-05-26 16:02:35.144 Ecowitt Gateway Connect attempt 2
2023-05-26 16:02:35.194 Error opening TCP port: Connection refused
2023-05-26 16:02:45.195 Ecowitt Gateway Connect attempt 3
2023-05-26 16:02:45.358 Error opening TCP port: Connection refused

as far as I can tell from the router logs etc it is not an issue with the connection but looks like it simply does not like connecting! usually after 3 or 4 attempts it will connect OK and continues for a few hours.

Anyone any ideas please?

Stuart

[Rover1822]

Yeah, this is a bit strange, as , correct me if I am wrong , but you have had WiFi issues now with multiple devices.

I have everything you have as far as GWs, and no issue. I can't speak to CMX as I don't use it.

I hate to say this, and please do not think I don't believe you have an issue, but the possibility exists the problem exists elsewhere.

Can you describe simply your network topology, router(s), mesh if applicable, access points if applicable etc. I know you have probably described all of these before, but as you know it is no fun to search the  forum.

- Rover

[broadstairs]

OK so I have 3 gateways now, a GW1000 and GW1100 both of which are working connected to my mesh system at 2.4ghz and currently working fine, the GW2000 is connected to my same mesh network at 2.4ghz and I can ping it with about a 5ms response however it refuses to talk to anything which tries to open a TCP port either CumulusMX or my own PHP script which gets the data via the same  API as both Weather Display or CumulusMX. I have no issue with anything else on my network at all, all phones,TVs tablets and my Weather Display using wifi to talk to the GW1100.

Right now the red light on the GW2000 is flashing slowly and the blue light flashes slowly. It does not respond to the phone app either and it has not reported now to Ecowitt for 54 minutes.

I cannot see why this is not working and cannot see any reason why anything on the network is an issue. My router shows the GW2000 as connected at the correct IP address as a statip IP.

Stuart

Just to add a power cycle has brought it back working again.

[kheller2]

Try a hard line for the GW2000... does that fix the problem?
There are a few discussions here that mesh networks have problems.

[SamiS]

OK so I have 3 gateways now, a GW1000 and GW1100 both of which are working connected to my mesh system at 2.4ghz and currently working fine, the GW2000 is connected to my same mesh network at 2.4ghz and I can ping it with about a 5ms response however it refuses to talk to anything which tries to open a TCP port either CumulusMX or my own PHP script which gets the data via the same  API as both Weather Display or CumulusMX. I have no issue with anything else on my network at all, all phones,TVs tablets and my Weather Display using wifi to talk to the GW1100.

Just to add a power cycle has brought it back working again.

Wifi issues are sometimes a real faultfinding nightmare.

Once upon a time I had a Zyxel adsl-wifi gateway and a Canon pixma inkjet connected to the wifi. The printer worked perfectly for days and weeks, until someone tried to print something from ipad. The result was half of the page printed, rest of it was white, and the printer dropped off from the wifi. All other devices on the network worked just fine despite the error, but the printer just could not get a connection anymore, no matter what you tried. The only way to get the printer back online was to reboot the zyxel itself. This repeatable issue was repaired after a firmware update from zyxel.

Mesh wifi also does have its own caveats. Sometimes the mesh systems try to control how the clients connect and to which accesspoint. Usually this is done by just kicking the client out from one ap so that it would connect to the another one. Usually the user doen not even notice, or there is only a small hiccup. But somethimes the clients are ”sticky” and for some reason do not want to change to the another ap. This may result headbanging, where the ap repeatedly kicks the device out, and it just keeps reconnecting to the same one. This issue can sometimes also be seen with band steering feature if 2.4Ghz and 5 Ghz ssid is the same.

A few years ago when completing my 5 day CWNA certification, our instructor said, ”It is surprising how well the wifi usually works despite the fact how horrible its design basics and backward compatibility features are”.

However just to be clear, I’m not saying that it must be a wifi issue you are experiencing. I’m just saying, that it is very hard to exclude that possibility except by connecting the GW2000 via wired lan.

—
Sami

[broadstairs]

Yes I know mesh systems can cause clients to jump access points however this system allows me to turn that off per client which is what I've done with all 3 gateways and since they are all connected to their nearest ap it should not be an issue. As I have already pointed out the GW2000 is connected and can be pinged on the correct IP, also my own script can create a socket but fails when it tries to read the firmware details via  the API using a x'50' command (CMD_READ_FIRMWARE_VERSION).

The problem starts with the gateway resetting the connection (Connection reset by peer) this is the client dropping the connection and then when trying to reopen the TCP port getting connection refused suggests the GW2000 is not listening despite being connected.

I can try the cable connection but that has failed in a similar way previously which is why I went to wifi.

Stuart

[broadstairs]

I just tried to get back to a lan connection but it refuses to work correctly, despite me setting a static IP and rebooting it gets a dhcp IP. I tried a hard reset (holding down the button for 5+ seconds) but it refuses to play ball.

I'm tired and fed up now so will try again in the morning.

Stuart

[SamiS]

On the earlier thread about gw2000 firmware 2.2.3 there were some users reporting issues with static address being lost on reboot and getting it back by reverting the setting to dhcp and again to static. Some also reported stability issues if both wifi and lan were enabled simultaneously.

If you get connection reset -error when querying the api, then most likely the server process on the gateway has indeed faulted. Since the problem has occurred both on lan and wifi, it is quite unlikely to be a network issue. Doing a packet capture might give some insight in what actually happens (eg. does the connection reset come from the gw or the client).

[broadstairs]

Well starting this morning refreshed and with a clear head I have it working now with a LAN connection. My issue with my LAN IP address assignment was my stupid BT Home Hub2, now sorted with a new IP and fixed in the DHCP bit in the Home Hub as static. Pings OK and now running OK in CumulusMX. My Home Hub does the DHCP assignment for both itself and the mesh system. I now have wifi disabled on the GW2000.

Now I'll wait to see if I get any glitches as it was taking a while previously to show up.

Stuart

[broadstairs]

Well after 24 hours the GW2000 has worked fine on the LAN connection, including a power cycle (to move PSU) and it reconnected OK, still unsure about wifi so I might try that in a few days time.

My BT FTTP Broadband contract is up later in the year and I will probably move suppliers as BT is expensive and get rid of the stupid Home Hub as it has quite a few design problems that way I can get a router of my own choosing which I cannot do right now because of Digital Voice.

Stuart