2 x RT-AX68U : AiMesh node constantly crashing

[fantom]

The main router is running fine, but the mesh node crashes more than once a day. Below are the first lines of the crash logs that I managed to save. Any ideas/suggestions to stabilize the mesh? Running the latest stable Merlin's release.

/tmp/home/root# head -10 /jffs/crashlog.log.*
==> /jffs/crashlog.log.1 <==
LOGFG80211-ERROR) wl_cfg80211_send_action_frame : Failed to send Action Frame(retry 6)
<1>Unable to handle kernel NULL pointer dereference at virtual address 00000008
<1>pgd = ffffffc0134a7000
<1>[00000008] *pgd=00000000134a1003, *pud=00000000134a1003, *pmd=00000000134a3003, *pte=0000000000000000
<0>Internal error: Oops: 96000047 [#1] PREEMPT SMP
<4>CPU: 0 PID: 975 Comm: eapd Tainted: P           O    4.1.52 #2
<4>Hardware name: Broadcom-v8A (DT)
<4>task: ffffffc01589b5c0 ti: ffffffc0134bc000 task.ti: ffffffc0134bc000
<4>PC is at __skb_recv_datagram+0x36c/0x498
<4>LR is at __skb_recv_datagram+0x6c/0x498

==> /jffs/crashlog.log.2 <==
LOG23: ffffffc013467cb8 x22: ffffffc013467be0
<4>x21: ffffffc013467cbc x20: ffffffc0134500a4
<4>x19: 0000000000000000 x18: 0000000000000000
<4>x17: 0000000000000000 x16: ffffffc00013ee80
<4>x15: 0000000000000000 x14: 000000000001537c
<4>x13: 00000000ff84c6f0 x12: 000000000002d028
<4>x11: 0000000000268420 x10: 0000000000000001
<4>x9 : 0000000000269410 x8 : 0000000000000000
<4>x7 : 0000000000268420 x6 : 0000000000000ff0
<4>x5 : ffffffc013467cbc x4 : 0000000000000000

==> /jffs/crashlog.log.3 <==
LOGr0: received packet on wl0.1.0 with own address as source address
<6>CFG80211-ERROR) wl_cfg80211_send_action_frame : Failed to send Action Frame(retry 6)
<6>br0: received packet on wl0.1.0 with own address as source address
<6>br0: received packet on wl0.1.0 with own address as source address
<6>CFG80211-ERROR) wl_cfg80211_send_action_frame : Failed to send Action Frame(retry 6)
<6>br0: received packet on wl1.1.0 with own address as source address
<6>br0: received packet on wl1.1.0 with own address as source address
<4>wl0: random key value:
<6>CFG80211-ERROR) wl_cfg80211_send_action_frame : Failed to send Action Frame(retry 6)
<1>Unable to handle kernel NULL pointer dereference at virtual address 00000008

==> /jffs/crashlog.log.4 <==
LOGgd = ffffffc0134aa000
<1>[00000008] *pgd=00000000134ae003, *pud=00000000134ae003, *pmd=00000000134a1003, *pte=0000000000000000
<0>Internal error: Oops: 96000047 [#1] PREEMPT SMP
<4>CPU: 1 PID: 976 Comm: eapd Tainted: P           O    4.1.52 #2
<4>Hardware name: Broadcom-v8A (DT)
<4>task: ffffffc01877c1c0 ti: ffffffc0134c4000 task.ti: ffffffc0134c4000
<4>PC is at __skb_recv_datagram+0x36c/0x498
<4>LR is at __skb_recv_datagram+0x6c/0x498
<4>pc : [<ffffffc0003ef9bc>] lr : [<ffffffc0003ef6bc>] pstate: 200001c5
<4>sp : ffffffc0134c7be0

 

[archiel]

I assume that by stable merlin release you mean 386.4. You might want to try changing the mesh node to use the default Asus firmware (do a factory reset after installing). Having merlin on the nodes does not add any significant benefits and in my case (AX58U node) mesh became much more stable. YMMV.

 

[pirx73]

I did not had crashes, but AiMesh was so wonky in 386.4 i rolled back to 386.3_2 which was rock stable for me, working without any issues since august. Also considering removing Merlin from the AiMesh node.

 

[bbunge]

If you must use Merlin go for the 386.5 beta. But for the node Asus firmware should work well.

 

[fantom]

If you must use Merlin go for the 386.5 beta. But for the node Asus firmware should work well.

Ok, but is not the point to have identical AiMesh code on all nodes? I do not have to use Merlin, but I read on this forum that is better to use the same hw/sw for AiMesh.

 

[OzarkEdge]

Ok, but is not the point to have identical AiMesh code on all nodes? I do not have to use Merlin, but I read on this forum that is better to use the same hw/sw for AiMesh.

If you want to use Asuswrt-Merlin, you may want to review its documentation: AiMesh · RMerl/asuswrt-merlin.ng Wiki · GitHub

OE

 

[fantom]

If you want to use Asuswrt-Merlin

This implies that there are cases when it is beneficial to run Asuswrt-Merlin on mesh nodes: do you mind sharing what those benefits could be? I just needed a dumb node and matching AiMesh implementations, but I may be missing other benefits.

 

[OzarkEdge]

This implies that there are cases when it is beneficial to run Asuswrt-Merlin on mesh nodes: do you mind sharing what those benefits could be? I just needed a dumb node and matching AiMesh implementations, but I may be missing other benefits.

My advice is to commission a new network with stock Asuswrt and use it for a while until you are fairly confident the new hardware is healthy.

Later, if you have a specific reason to use Asuswrt-Merlin, install it on the router/root node only as is recommended.

To better understand this advice, review the Asuswrt-Merlin documentation. In short, it is a platform for doing much more with your network... right now you are just trying to commission a new network and don't need to be going down rabbit holes, imo.

OE

 

[fantom]

Yeah, I will need Asuswrt-Merlin firmware. I interpreted "If you want to use Asuswrt-Merlin" with "on the node". If there is no custom functionality running on the nodes, then I misinterpreted what you said.

 

[OzarkEdge]

Yeah, I will need Asuswrt-Merlin firmware. I interpreted "If you want to use Asuswrt-Merlin" with "on the node". If there is no custom functionality running on the nodes, then I misinterpreted what you said.

I think you should stop dissecting posts and trying to read between the lines and just read the damn documentation.

OE

 

[fantom]

I think you should stop dissecting posts and trying to read between the lines and just read the damn documentation.

OE

Thx. I just did so and went to the latest Asus firmware on the node, but I also went back to WPA2 from WPA2/WPA3 and there was no crash in over 12 hours; could be related.

 

[OzarkEdge]

Thx. I just did so and went to the latest Asus firmware on the node, but I also went back to WPA2 from WPA2/WPA3 and there was no crash in over 12 hours; could be related.

WPA2/WPA3 is working here with my few clients... the one AX connects with WPA3, the ACs connect with WPA2. It could be a picky client or one that needs a driver update; or it could be a firmware glitch that will have to wait for a fix.

OE