New Communications Issues..

[rrepic]

Hi Michael,
Communications using the GC WiFi to Rain8Net Pro have been working ok but not without troubles. They range from having to reset my wifi router to strange responses. I am going to convert off wifi and try wired ethernet.

I have upgraded to 2.11 and reset my wifi again and communications seems to be working. When I manually turn on a zone, it works fine. Today I noticed that my scheduled zones were actually running (from mcs view) but no valvles were actually turned on. From my HS log:

5/30/2011 10:22:00 AM mcsSprinklers Rain8NetSend Line 390 Unable to write data to the transport connection: An established connection was aborted by the software in your host machine.

5/30/2011 10:22:05 AM mcsSprinklers OpenRain8NetPort Line 30 Access to the port 'COM3' is denied.

5/30/2011 10:23:03 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Cannot access a disposed object.Object name: 'System.Net.Sockets.Socket'.

5/30/2011 10:32:02 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:32:05 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Object reference not set to an instance of an object.

5/30/2011 10:32:05 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:33:02 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Cannot access a disposed object.Object name: 'System.Net.Sockets.Socket'.

5/30/2011 10:33:04 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Cannot access a disposed object.Object name: 'System.Net.Sockets.Socket'.

5/30/2011 10:34:02 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:34:05 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:34:13 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Cannot access a disposed object.Object name: 'System.Net.Sockets.Socket'.

5/30/2011 10:34:13 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:38:02 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

5/30/2011 10:38:11 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 103 Cannot access a disposed object.Object name: 'System.Net.Sockets.Socket'.

5/30/2011 10:38:11 AM mcsSprinklers OpenRain8Net_Socket 172.30.0.16:4999 Line 106 Object reference not set to an instance of an object.

When I check general status or zone touch status, mcs thinks its in the middle of running a normal program.


Thought I would check with you first on the communications side to see if you have seen this error before.

Thanks as always!

-Ray

mcsSprinklersP 2.11

[Michael McSharry]

The line 103 and 106 are expected when mcsSprinklers gives up on an existing connection and tries to start a new one. In latest version I have demoted these to debug messages rather than normal log messages. They occur because the connections are intentionally removed, but there are still active threads that will be listening or completing their transaction and the intentionally destroyed object is not longer available.

The one about COM3 is not expected. The line 390 one is normal when the WiFi/serial devices drops an existing connection.

A have been using a little ASUS access point for my WiFi testing with either a Lantronix EPS or Quatech IP/Serial devices. I have never seen any network issues until I move the access point out of WiFi range. For testing I typically either remove power to the IP/Serial or access point or remove the cat5 cable between the two.

[rrepic]

Interesting...

The com3 error isexpected as its my rain8net that is on serial control. I still have it in my config although I dont use it as Ive replaced it with the rain8netpro. Ill remove it now that the rain8netpro is operational.

As for ip coms, I reset the global cache wifi-serial adapter (pulled the power plug). It came back on the network then for some reason ALL valves turned on the rain8netpro. I manually turned all valves off on the rain8netpro as mcs didnt show all valves on. Strange.

I am going to shut everything down and bring it back up. I sense my wifi is flaky and will concentrate efforts there.

Let me know if you concur.

Thanks Michael!!!

-Ray

[Michael McSharry]

You should pay attention to the low level serial communications between the PC and Rain8NetPRro. The Rain8NetPRO should not power up in a active configuration unless commanded via a 3-byte sequence that meets protocol. In hex this would be 31 XX FF where XX is the module ID.

As I indicated before I have no issues with WiFi connections. I also do not have the GC hardware. In general I would debug from the high level protocol down toward the low level one. Your concept of going wired rather than WiFi is a good one. At the same time observe the serial commands from the PC to confirm you are commanding what is expected. mcsSprinklers would send an all-on command at startup if it thinks there is a schedule to do such. This is not likely. On startup mcsSprinklers does a status request and set the valve devices to correspond to the physical reported status.

The general debug log contains communication and scheduling information so keep it enabled while you are investigating.

[gerenb]

I have the same configuration and most of the communications problems were resolved in the latest versions. I have noticed, however, that once in a while the plug-in will give up attempting to communicate with the Global Cache unit after a few minutes. Once in this state, it will never try to reconnect - even when the communications have been restored and I can ping the device.
My solution has been to automatically restart HS on a daily basis 30 minutes before a possible watering event. This virtually guarantees the watering event will occur. IMHO, the plug-in should never completely give up trying to reconnect.

[Michael McSharry]

The design is to retry and reset and retry for a fixed number of time for the current command and then finally discard the . The process should repeat for the next command that is delivered so if the WiFi device has reconnected by then is should proceed as normal.

There was also the reset management added in recent versions where if mcsSprinklers have four consecutive failures to connect it will cycle power on the reset device. You may want to try this as well to see how recovery behaves

[Mael]

There was also the reset management added in recent versions where if mcsSprinklers have four consecutive failures to connect it will cycle power on the reset device. You may want to try this as well to see how recovery behaves

Is there a setting in version 2.11.0.13 to activate this or is it built in?

[Michael McSharry]

I should be in the are were you define your Rain8Net parameters on the Rain8/Devices page and will include the device code and the delay following power cycle for the unit to go through its initialization.

[Michael McSharry]

I would like to get to a stable approach for dealing with connection loss with the WiFi or other TCP adapter. The philosophy that is implemented now is to try multiple time, reset the adapter if setup to do that and then retry multiple times again. After this the command that was being sent is discarded and the process starts again with the next command.

If the Rain8 Page setup is poll evey minute then effectively the attempt to regain connection is relatively continous. If it is setup to request status after command then a lost connection recovery will not be attempted until there is a need to send another command.

I think there is a mixture of users that wanted a reset on the WiFi/Serial device and others who found that a HS restart is what worked. If your setup is still having problems then I am receptive to suggestions as to what mcsSprinklers should do to recover in your specific situation.