I purchased this router a week ago and installed LEDE 17.01.4 on it when I received it. All was well for the week until today. I came home to both (2.4/5g) Wifi not broadcasting, no internet connectivity on the wired computers and unable to access the router's web interface. It seemed to be able to redirect from the IP address to /cgi-bin/luci but no further. Power light was on and wired network lights flashing but wireless lights off.
I power cycled and everything came back as normal, what's the best way to determine the cause in this case when the router is still powered on and semi functional but the web UI is not responding?
All the logs I could find in the web UI seem to reset when the router is rebooted.
ssh into device and check dmesg / logread, You might consider sending logs down the wire to a syslog instance, or logging to a usb device. If you have not done so, I would suggest pre-built until an update finds its way into a stable release.
@anomeome's great suggestion about a downstream syslog instance should be considered, as it's the best way to ensure one isn't hindered by a device reboot. A syslog server also comes in extremely handy if you're running multiple BSD/Linux systems, providing a centralized database for all devices' system logs.
In general, it's better for users to be sent to a centralized place, in this case for the Marvell WiFi drivers, rather than a thread. The information regarding @kaloz's repo and how to build have always been in the WRT AC Series Wiki's 88W8864 mwlwifi section
. @Cube00 The only way to troubleshoot something like what you experienced is to either SSH in (if possible), otherwise you'll need to utilize the serial header to copy the log to a text file and either save it to /root/sys.log via logread > /root/sys.log or save it to mounted USB storage.
There's no way to determine whether this was WiFi related or not, and what you describe (if you also didn't have WAN access), has occurred on other WRT AC Series routers due to a kernel panic, WiFi, and a handful of other things, but there's no way to discern which without the system log.
Thanks for the replies so far. I upgraded the wifi kmod as suggested to Package: kmod-mwlwifi-4.4.92+10.3.4.0-20171214-843d00c-9322
I mounted a USB drive and set the logging to a file on there.
Crash happened again, nothing in the system log on the USB drive but info level logging, no errors.
SSH was inaccessible along with everything else (wifi down, WAN down, LAN still up allowing access to other machines on the wired network).
However this time when powering off and on and it still didn't come up.
Took two further on/off cycles for it to come back up.
Is the only option now a serial connection? Given it's a new router will opening it to do this void the warranty? Should I just give up and go back to stock to rule out a hardware problem?
What exactly do you believe is wrong with the hardware? There isn't an issue with the hardware, as this is a software issue, and as such, recommending anyone simply return their product based upon a 3rd party software issue is simply not an intelligent suggestion.
You have a corrupted firmware version on one of your partitions, so build a new image and flash it via serial to both partitions.
LEDE 17.01.4 is using kernel version 4.4, which could be the cause... I've been using 4.9 on my ACS and I think I've experienced a random crash maybe 3 times in over 2 years. You can only get kernel 4.9.x via the development builds or by building your own image (something I recommend all to do if they're running LEDE, as it's far more convenient for the user).
Kernel 4.4 has several issues on the WRT AC Series routers, so even if it's not the cause of this specific issue, it's still recommended to utilize 4.9
so you cycled the power 4 times, 3 failed boots, and back to the other partition, but is that OEM or LEDE.
might be a good idea to flash OEM to one partition if not already there, and just flash over the other partition repeatedly until you resolve issue.
seems like a bad idea to consider flashing the same, possibly borked image, to both partitions, removes your failsafe
iirc popping the top does void your warranty, so be sure the unit is good first.
I am not aware of any issue with 4.4k on any members of the wrtpac, so would expect 17.01.4 to work fine.
there have been reports of certain clients causing a reboot, but iirc there was some logging to be had, so your description varies, report mwlwifi issues
are you sure as to validity of downloaded image, CRC check against what is published in download area
Wasn't there issues with random reboots on 4.4, along with odhcpd registering static IPs with a netmask of /32, or did those get resolved?
...And no, it does not void the warranty if you disassemble the case. The only thing that would be affected would be Linksys tech support, as they'll only offer tech support if one is running OEM firmware. In the unlikely event something is hardware related, one only need to flash OEM firmware, contact Tech Support, at which point they'll likely provide the user with a debug firmware image which will need to be flashed.
In the US at least, consumers are protected under the Magnuson–Moss Warranty Act of '75 (15 USC § 2301), with the burden of proof residing on the manufacturer to prove what a consumer did, or did not do, is the cause of whatever arbitrary issue is at hand (this is also why rooting one's phone does not void the warranty)
You might be able to pop the top on the new wrt devices without voiding the warranty, my wrt1200 v1 didn't have a warranty sticker or breakaway tabs, so unless linksys is paying attention to the torque of the screws or you have to solder i'm not sure how they'd know, that obviously is no guarantee other wrt devices are the same though, so attempt at your own risk.