Archer C7v2 loses LAN connectivity randomly

[TopDog]

I’m running 18.06.1 OpenWRT on a TP-Link C7v2, configured to be just an AP (no DHCP, one LAN port bridges rest of home into br-LAN for WiFi and one locally connected laptop)

I was logged in via ssh checking some logs and settings, then asked it to reboot with a reboot;exit command. When it came back, it had no LAN. I could connect via WiFi and see the router was up. But no LAN, as my hardwired laptop could not see the internet, I could not ping the router.

A cold boot took care of it.

Then later, it lost the LAN again, I could get in via wifi, but nothing in the logs looked abnormal. So I went digging around for possible references to this to see if I could find a solution.

It looks like this is a problem that has affected several TP-link boxes using some common QCA SoCs.

Here are the related threads I found:

This one indicated that high traffic caused the drop.

Weird Internet disappearance on Archer C7v2

Are you running any applications on your router aside from the default install?

This thread, while about a sister TP-Link model WR1043ND, the SoC (qca956x and qca953x) addressed is pretty close to the one in the C7v2 (QCA9558), and a C7 user reports seeing the exact same symptoms in that thread:

LAN stops working every now and then

I know, and the replacement unit they got in return suffers from the same issue as I heard. Also I sent them the links for this forum threads, and the one I opened on tp-link forums: http://forum.tp-link.com/showthread.php?97391-TL-WR1043ND-v4-LAN-WAN-ports-stop-working For now I take my time with the return, maybe we/they can find a solution, or until they come out with v4-rev.2 hardware (which I probably will get as a replacement then ). Btw I'm from Hungary. I really don't …

The associated comments in a bug report referenced in that thread sure makes it sound like there is a high probability of the C7v2 experiencing a similar thing, as its SoC (QCA9558) is a close relative. The bug report is: https://bugs.openwrt.org/index.php?do=details&task_id=762

The related fix focuses on small packets flood, which correlates in my case to coming back up and seeing a flood of mDNS and small UDP traffic from a variety of IoT devices connected to this AP.

So my question is whether this bug affecting C7v2’s (and possibly other systems with that SoC), already has an open bug report? I looked, but did not immediately find one.

Anyone else being affected by this?

[tmomas]

Does this also happen with a snapshot image?

Sounds like this commit could be related.

[slh]

At least using snapshots (as in binary builds without changes) won't have any effect here, as enable_sgmii_fixup is only set for a single device - the AVM Fritz!WLAN Repeater 450E.

[JonFo]

I've also observed this behavior, but it's been so random, I rarely get to capture any info. I confirm that once the LAN goes down, I can get in via WiFi and the UI and ssh respond, just nothing via the switch.

After a reboot, it starts working again. I have many devices and a busy network, so really hard to correlate to any specific activity.

This has occurred with both 17.01.x builds as well as with the latest 18.06.1, maybe four or five occurrences across 14+ months.

[TopDog]

I have not tried snapshots yet, but as slh mentions, it looks like it would have the same issue.

That commit does look relevant, thanks. I don't do builds, so no idea on how to test.

[wjwj]

@TopDog: Would you be so kind, read my post eth0: link down and cross-check your logs, whether they show similarities?

[TopDog]

wjwj, will do next time it drops, I did not manage to grab the logs the last time, but IIRC, they did not have the entries your logs show.