Network problem on control network B
We are facing some network trouble.
Secondary Controller network keep going down.
In Moxa switch overview all switches in network ring with controller connected becomes red. Switches only connected to CS servers is still up.
In RNRP all controllers are being disconnected and connected again for aprox 15-50 minutes.
This happens a couple of times each day.
Please see attachment for more info.
Not only the controllers become disconnected, the Moxa switches seem to have issues themselves.
Does these Moxa switches have any internal log that can be pulled and checked?
Is swittch "Uptime" as expected (>1day) or do they crash/reboot (<1 day)?
I suggest checking the firmware of the Moxa switches and if there are any documents on Moxa's web containing known issues and corrections/patches.
Its also true that ABB recommend using fixed speed and duplex over auto negotiation, but as far as I know, the current PM891 firmware has a flaw that may cause a port set for fixed speed & duplex to become with auto negotation if certain timing of events take place when powering up the PM891. In this situation, a switch set for 100 FULL would force the controller to fall back to 100 HALF (the auto negotiation ruleset dictates, if no response on duplex negotation is received, fall back to HALF).
Until this problem has been corrected (check release notes) I recommend keeping PM891 ports at auto-negotiate (and of course also in the switch).
If I interpret the MXView management software correct, the connection with two switches (172.16.81.249 and 172.16.81.252) were lost for about a minute around 2015-05-27 14:24.
This seem odd to me and should be investigated further; a minute could very well be the time a switch crash / restart would take to recover from.
I asked for logs from these switches, but you sent pictures from another switch instead (172.16.81.245). Please probe the switches reported lost by MXView and check if they indicate a crash (e.g. uptime no longer than what has passed since the last loss of connection from MXview) or anything else that could explain the loss.
I also recommend that you get the clock synchronized in the switches so that you can look at absolute time rather than relative time when comparing "RNRP down/up" with switch log information. You can read the time of any AC800 controller or connectivity server within range. Win32Time can be started in *server* mode (client mode should be disabled to avoid the PC from becoming slave to more than one protocol - required settings are listed in the 800xA Automation Network User's Guide)
I have reset logs from switches and configured them to synchronize with controllers.
Then I waited till the nodes started go down again.
Switch 172.17.81.246 report of "Port link 7 down".
Switch 172.17.81.245 report of "Topology changed"
It seems to me that there is something wrong with switch 246 and since 245 is ring master he will open standby port.
Unfortunately the timestamp from MXview is not synced and I dont see why..