Started okay, fine for a week, then nothing

[estragon]

Setup:
Linksys wrt54g flashed with dd-wrt running in client bridge mode.
3 Classics w/ethernet to bridge
WiFi bridge to Sierra Wireless cell access point.

MyMidnite seemed fine for about 6 days after setup.  We turned off the access point for the night as we generally do, and the next day and thereafter (about 2 weeks), nada.

I've messed about with the bridge to no avail, and also tried bridging to a cellphone set up as an access point. 

I can put a windoz laptop on the wired side of the bridge and all seems fine.  Can ping mymidnite.com etc. and also the classics.

LocalApp works fine both on the wired side of the bridge and the WiFi side, and so does Graham's classic smartphone app (very nice BTW).

Tried sniffing packets with wireshark, but all I'm seeing is periodic UDP packets.  The bridge lan ports are likely switched though, so wouldn't see much else anyway.  Tried putting the classics on a hub with the laptop, but that just lit up the collision light on the hub.  Do the classics have auto-sensing ethernet?  Maybe I need to make up a x-over cable.

I've tried static ips on the classics and also putting in the alternate dns for mymidnite.  Power cycled etc.  Still no joy.

I don't suppose there's a little command line interface in the classic for debugging is there?

Anyway, I've seen a few other threads with similar symptoms but didn't see a resolution.  Any ideas?

[atop8918]

If you are turning off your access point at night you will also need to enable the A-RST or auto reset function on the Classic.
You can also try simply resetting the Classic using the Local App: Config->Basic->Bully Menu->Reboot.

[estragon]

Thanks for your reply.

I do have auto rset enabled.   Have also done the bully menu reboot (many times) and also power cycled. 

Is there a back-off algo built into to classic or does it try  exactly once to connect every 10 minutes and then stop after n tries?  I'm guessing the latter.   If so do you know the value of 'n'?  Also, from your reply I assume a bully reset should set the value of 'n' to 0 and the classic to again make 'n' tries over n(10) mins.

The really strange thing is it looks like all was well for a week or so, then nothing. 

[atop8918]

There is no back-off algorithm as such. The unit will attempt to connect once every 5-10 minutes depending on the firmware you are running. The server will either let the unit authenticate and deliver data or will kick it off if it thinks that not enough time has passed between samples. If there are any other connection errors then the Classic will attempt to reconnect after a short pause until it explicitly hears from the server.

With the Classic, there is some type of network traffic that we have been unable to identify that seems to lock up the remote access portion of the stack. A bully reset simply resets the stack to its initial state and clears out all the state machines so any accumulated errors are removed. The current protocol has been deprecated in favor of a much lighter-weight protocol that is much more error tolerant than the current system. We will be releasing this firmware to coincide with the release of the new mymidnite website.

In the meantime if you have reset the Classic and are still not getting mymidnite data, you may want to check for IP collisions on the network unless you've got everything set up with static IP addresses.