Losing Network and other hardware on rack every 20 days or so

[FrankFlintstone16044]

Hi

I have an RV340 router dual WAN input.  WAN 1 is 4G Archer router and WAN 2 is a BT fixed line. 

The network is frequently lost, it goes every 20 days or so and it locks WAN 1 completely. We have now installed a Powerbar to autoping WAN 1 and then reboot the Archer.  As WAN 1 is 4G I expect some instability. 

The Switch is 28 port TPlink small business switch with RST configured.

BUT we are also losing connectivity to other parts of the rack which runs a the home automation.  We are losing the AV system, we lose the CCTV system. 

Whenever we lose an item of hardware rebooting the individual hardware does not bring it back, the entire rack needs to be rebooted.  

Can anyone help?

[marce1000]

 - In general it then becomes advisable to configure a syslog-server for all involved Cisco devices to capture logs at a central point, use the same syslog-server for other equipment too, if they support the syslog-protocol. Also when this is done, observe logging during normal operations too because sometimes messages could be logged which are 'ahead' and or indicators of a more serious problem coming up.

 M.

[nagrajk1969]

Hello Frank

I would strongly suggest that you apply the below step, and do once a apply-permanent save and reboot once after this change

1.Under WAN/MultiWAN gui page, both WAN1 and WAN2 have the attached setting of "Enable Network Service Detection" by default.

2. I strongsly suggest that you please "DISABLE" this setting for both wan1 and wan2 interfaces...and then apply-save and then do a reboot once after this

Check this out please

thanks

regards

[FrankFlintstone16044]

HI

Thank for the advice.  I have no issue in doing this.  Can you explain why I should do this and why it will make the issue any better please?  Not doubting the solution just trying to understand what you have proposed.

Thanks

Frank

[nagrajk1969]

Yes sure the points of explanation are as below:

1. This is Dual-WAN router with Fail-Over and Load-Balancing modes supported. So in order support say fail-over for example, there is the service "MultiWAN" which is used

a) Now when both wan1 and wan2 are confgured and active/UP, the mutiwan service will keep track of the status of the link on wan1 and wan2 by sending some keepalives traffic to a destination on each wan interface. The keepalives are PINGs - 3 pings every 5 seconds or something like that 

b) The keepalive is enabled by the setting "Enable Network Service Detection" - which is always enabled by default for Static-Interfaces.

please Note it is NOT enabled either by default AND SHOULD NOT BE ENABLED FOR PPPOE-WAN & USB/3G/4G-WAN interfaces... 

- this setting is there separately for wan1 and wan2 interface under MultiWAN

Note-1: By default the enable-network-service-detection is set to "Default-Gateway" of each WAN interface...

Note-2: The user may set to send the pings to some other remote site which will be assumed to always be UP and always send the Ping-Replies...like the google dns-server 8.8.8.8....etc

c) Now what happens is that sometimes due to load on wan1/wan2, or simply becos the "default-gateway" is NOT replying to the ping-requests, or maybe for some reason, even if there is no load on the wan, some of the ping-requests sent (5 pings) only 3 replies are being recieved or just 1, or none at all....

- all or any of this above issues will trigger the multiwan service to make the WAN1/WAN2 interface "offline", meaning it will stop forwarding traffic on the said wan interface and in case of fail-over mode start re-directing traffic to the other wan interface, if its active

- So as long as the wan interface is in offline mode, no traffic will be forwarded thru the wan interface

 2. Hence my suggestion was simply stop using the "Enable Network Service Detection" on both WANs and this will ensure that both wan interfaces are always ONLINE

- In this case, the multi-wan service will still re-direct the traffic to other wan interface (for fail-over) ONLY if the said WAN interface's link is down or the network service on that link(ethernet-interface) is DOWN for any reason whatsoever..

Hence my suggestion...that under heavy load conditions, even if you set the "network service detection" to 8.8.8.8 (or to default-gateway)....Ping traffic with their small timeouts tend to get easily dropped somewhere in the network...and hence replies will not come...this will result in the multiwan service bringing the wan interface down (offline)....and after all pings are getting replied, it will bring up the wan interface (online) again...this will be kind of network-flapping, 

- so why not disable the "network service detection" altogether...let the wan interfaces be always online..and therefore UP and active

Also there is this additional justification for keeping the network-service-detection disabled....its NOT supported or should not be enabled for PPPoE-WAN interfaces right....so how does it make sense to enable it for Static-wan interfaces????...so try it out...it may solve your existing problem and make it stable

Also one more point...have you updated the firmware on the RV34X to the latest firmware v1.0.03.21???? Update it

[FrankFlintstone16044]

Hi

Thanks for the detailed response.  Before I try it can you help me understand what you said here "it is NOT enabled either by default AND SHOULD NOT BE ENABLED FOR PPPOE-WAN & USB/3G/4G-WAN interfaces."

Specifically this part..."SHOULD NOT BE ENABLED FOR 3G/4G-WAN interfaces".   Where did you get this from?  I have never seen anything that says the network detection does not work over 4G but it is entirely feasible!  Can you tell me the source of the data, the client will be as interested as I am.  Thanks again for the help its great.