RT-AC88U loses WAN and LAN access, only WiFi still remains until physical replugging of ports?

[ninja888]

Hi Guys,

I’m seeing a very strange problem with my Asus RT-AC88U (at the time running Merlin 384.10_2 but I’ve since upgraded firmware to 384.11 Beta 1) where, randomly (it happened about 4 times in the last couple of weeks) , the router will lose WAN connection (WAN light on router is RED) and the router will not be able to ping ANY connected LAN devices. The only way I can access the router GUI / SSH is via local WiFi which seems to be unaffected…

When this happens, the only way to recover the router is to physically remove all of the LAN and WAN connections, replug the WAN connection and then replug all the LAN connections… I’ve tried a rebooting the router when this issue occurs but the router simply reboots to the same state (WAN/LAN ports down and only WiFi access) …

Questions:
(1) Any idea why this might be occurring and anything that can be done to prevent it?
(2) Once this occurs, is there any commands to virtually unplug/replug the WAN/LAN devices rather than having to physically re-plug them?


While the router was experiencing this issue, I ran the following command over SSH via WiFi:

admin@RT-AC88U:/tmp/home/root# robocfg show
Switch: enabled
Port 0: DOWN enabled stp: none vlan: 1 jumbo: off mac: 00:00:00:00:00:00
Port 1: 1000FD enabled stp: none vlan: 1 jumbo: off mac: <ROUTER_PORT1_MAC>
Port 2: 1000FD enabled stp: none vlan: 1 jumbo: off mac: <ROUTER_PORT2_MAC>
Port 3: 1000FD enabled stp: none vlan: 1 jumbo: off mac: <ROUTER_PORT3_MAC>
Port 4: 1000FD enabled stp: none vlan: 2 jumbo: off mac: <ROUTER_PORT4_MAC>
Port 5: 1000FD enabled stp: none vlan: 1 jumbo: off mac: <ROUTER_PORT5_MAC>
Port 6 <MISSING INFO FROM ROBOCFG SHOW?????>
Port 7: 1000FD enabled stp: none vlan: 1 jumbo: off mac: <ROUTER_PORT7_MAC>
Port 8: 1000FD enabled stp: none vlan: 2 jumbo: off mac: <ROUTER_PORT8_MAC>
VLANs: BCM5301x enabled mac_check mac_hash
1: vlan1: 0 1 2 3 5 7 8t
2: vlan2: 4 8u


The physical connections to the router were as follows:

WAN Port = WAN Cable to Fibre Modem
LAN Port 1 & 2 = Link Aggregation to NAS
LAN Port 3 = SERVER1
LAN Port 4 = <EMPTY>
LAN Port 5 = SERVER2
LAN Port 6 = <EMPTY>
LAN Port 7 = SERVER3
LAN Port 8 = <EMPTY>

Some initial observations:
- LAN Port 6 is not shown in the “robocfg show” output?
- GUI shows Port 7 operating at 100Mbps (which is correct) but “robocfg show” shows it operating at 1000FD (which is incorrect)?

Steps taken next:
(A) Rebooted router – Issue persisted so removed WAN cable – LAN Port 4 is then shown as DOWN in “robocfg show” but nothing is connected to LAN Port 4 – WAN link to modem was connected to Port 0?
(B) Removed Cable 7 and Cable 5 but Port 5 and 7 still shows as UP in “robocfg show”
(C) Removed Cable 3 - Port 3 still shows as UP But Port 1 goes DOWN in “robocfg show” - As soon as removed Cable 3, WAN is restored (WAN light almost straight away goes white) but still cant ping the remaining LAN devices (Port 1 / 2 bond to NAS). Only after rebooting router can it ping Port 1 / 2 LAN device. Plugging each other LAN cable back in it can ping LAN Device on Port

Router syslog output:
Issue occurred around May 1 14:48:00 but the following is only output to the syslog after the event occurred:

May 1 14:51:26 WLCEVENTD: eth2: Disassoc <ROUTER_PORT7_MAC>
May 1 14:51:27 WLCEVENTD: eth2: Assoc <ROUTER_PORT7_MAC>
May 1 14:51:38 ovpn-client1[3072]: [TG-OVPN-CA] Inactivity timeout (--ping-restart), restarting
May 1 14:51:38 ovpn-client1[3072]: SIGUSR1[soft,ping-restart] received, process restarting
May 1 14:51:38 ovpn-client1[3072]: Restart pause, 5 second(s)
May 1 14:51:43 ovpn-client1[3072]: NOTE: the current --script-security setting may allow this configuration to call user-defined scripts
May 1 14:51:43 ovpn-client1[3072]: TCP/UDP: Preserving recently used remote address: [AF_INET]<VPN_IP>:995
May 1 14:51:43 ovpn-client1[3072]: Socket Buffers: R=[122880->122880] S=[122880->122880]
May 1 14:51:43 ovpn-client1[3072]: UDP link local: (not bound)
May 1 14:51:43 ovpn-client1[3072]: UDP link remote: [AF_INET]<VPN_IP>:995
May 1 14:51:53 WLCEVENTD: eth2: Disassoc <ROUTER_PORT7_MAC>
May 1 14:51:55 WLCEVENTD: eth2: Assoc <ROUTER_PORT7_MAC>
May 1 14:51:56 WLCEVENTD: eth2: Disassoc <UNKNOWN_MAC_ASSUME_PORT6?>
May 1 14:51:59 WLCEVENTD: eth2: Assoc <UNKNOWN_MAC_ASSUME_PORT6?>
May 1 14:52:13 ovpn-client1[3072]: [UNDEF] Inactivity timeout (--ping-restart), restarting
May 1 14:52:13 ovpn-client1[3072]: SIGUSR1[soft,ping-restart] received, process restarting
May 1 14:52:13 ovpn-client1[3072]: Restart pause, 5 second(s)
May 1 14:52:18 ovpn-client1[3072]: NOTE: the current --script-security setting may allow this configuration to call user-defined scripts
May 1 14:52:21 WLCEVENTD: eth2: Disassoc <ROUTER_PORT7_MAC>
May 1 14:52:22 WLCEVENTD: eth2: Assoc <ROUTER_PORT7_MAC>
May 1 14:52:24 pppd[2474]: Serial link appears to be disconnected.
May 1 14:52:24 miniupnpd[23701]: Failed to get IP for interface ppp0
May 1 14:52:24 miniupnpd[23701]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
May 1 14:52:27 WAN_Connection: Fail to connect with some issues.
May 1 14:52:27 DualWAN: skip single wan wan_led_control - WANRED off
May 1 14:52:27 nat: apply redirect rules
May 1 14:52:28 ovpn-client1[3072]: RESOLVE: Cannot resolve host address: <VPN_URL>:995 (Name or service not known)
May 1 14:52:28 ovpn-client1[3072]: RESOLVE: Cannot resolve host address: <VPN_URL>:995 (Name or service not known)
May 1 14:52:28 ovpn-client1[3072]: Could not determine IPv4/IPv6 protocol
May 1 14:52:28 ovpn-client1[3072]: SIGUSR1[soft,init_instance] received, process restarting
May 1 14:52:28 ovpn-client1[3072]: Restart pause, 5 second(s)
May 1 14:52:30 pppd[2474]: Connection terminated.
May 1 14:52:30 pppd[2474]: Modem hangup
May 1 14:52:33 ovpn-client1[3072]: NOTE: the current --script-security setting may allow this configuration to call user-defined scripts
May 1 14:52:33 ovpn-client1[3072]: RESOLVE: Cannot resolve host address: <VPN_URL>:995 (Name or service not known)
May 1 14:52:33 ovpn-client1[3072]: RESOLVE: Cannot resolve host address: <VPN_URL>:995 (Name or service not known)
May 1 14:52:33 ovpn-client1[3072]: Could not determine IPv4/IPv6 protocol
May 1 14:52:33 ovpn-client1[3072]: SIGUSR1[soft,init_instance] received, process restarting
May 1 14:52:33 ovpn-client1[3072]: Restart pause, 5 second(s)

 

[Bamsefar]

For testing purpose, and verify a known thing with the LAN port 5-8, could you remove link aggregation to free up one LAN port and then connect all devices on LAN port 1-4. Any difference? I have problems of all kinds if I use port 5-8 on my AC88 (which more or less makes the AC88 useless if one can not use thoose ports...).

Then make sure the fibermodem don't have problem with the GBIC / SFP (depending on what is used inside the fibermodem). I had serious problems with my SFP and when I changed to another make of that SFP everything just worked.

 

[ColinTaylor]

Some initial observations:
- LAN Port 6 is not shown in the “robocfg show” output?
- GUI shows Port 7 operating at 100Mbps (which is correct) but “robocfg show” shows it operating at 1000FD (which is incorrect)?

That's normal. Only 4 of the LAN ports are on a Broadcom chip, the other 4 are on a separate Realtek chip. So the robocfg output is misleading.

https://www.snbforums.com/threads/odd-rt-ac88u-port-assignments.40744/

 

[ninja888]

Intriguing!

If robocfg can't be trusted to output the correct config for both the Broadcom (Ports 1-4 and WAN?) and Realtek (Port 5 -8 ) switches, is here a command line command that can be trusted and allows the user to disable / re-enable the switches/ports individually to mimic physical re-plugging individual cables (without actually being there since I may want to do this remotely via a secondary access point to the network)?

I assume there at least must be a command which lists the correct port config states across both switches given that the main homepage in the Web GUI (System Status -> Ethernet Ports) appears to list the correct state/speed of the WAN and LAN1-8 ports?

 

[ColinTaylor]

I assume there at least must be a command which lists the correct port config states across both switches given that the main homepage in the Web GUI (System Status -> Ethernet Ports) appears to list the correct state/speed of the WAN and LAN1-8 ports?

I don't know if there's a userspace command. The GUI queries the Realtek chip directly.

 

[ninja888]

I don't know if there's a userspace command. The GUI queries the Realtek chip directly.

Ok thanks Colin...

If there’s no userspace command to view the reliable states of the ports across both switches, are you aware of any command which can reliably disable / reenable the interfaces / ports / switches to mimic a physical replugging of each of the cables? This would at least in theory help me to revive the router remotely when / if this issue occurs again...

Alternatively, I’m hearing both here and in other historic posts that it maybe best to just abandon using Ports 5-8 on this router and instead using a separate chained switch....

 

[ColinTaylor]

I suggest in the first instance you disable link aggregation and run like that to determine whether that is the cause.

 

[ninja888]

Yes, I will disable link aggregation and just use the first 4 LAN Ports to see if that makes any difference but it would still be good to know if anyone has any suggestions for userspace commands which can be run over local SSH (WiFi) to disable/enable the ports in case this happens again - Especially since the router is remote and getting to the router and physical replugging of the cables is a chore!

Thanks