384.5 AC-88U dies after 14 day uptime

[Morac]

I have no idea why, but once my AC88U running Merlin 384.5 hits about 14 days uptime, the router just locks up and won’t respond and needs to be hard reset. That just happened today. I checked the router web page this morning and it had 293 MB free (memory slowly goes down over time, but not enough to run out) and then a few hours later I started getting brief Internet outages and finally no network activity at all in the middle of playing a network game on my iPad.

I couldn’t connect to Wifi after disconnecting, anything connected to Wifi or Ethernet couldn’t access the router, LAN or WAN. Everything was just dead. I power-cycled and everything was fine again. There was nothing in the logs. I never had issues prior to 384.5. I ran 384.4 for months without issue.

Any idea what’s going on?

 

[Clark Griswald]

I can recommend the latest alpha2 for the 88. Mine has been running smoothly since install.
https://www.snbforums.com/threads/p...merlin-384-6-test-builds-are-available.47385/

 

[Xentrk]

I have no idea why, but once my AC88U running Merlin 384.5 hits about 14 days uptime, the router just locks up and won’t respond and needs to be hard reset. That just happened today. I checked the router web page this morning and it had 293 MB free (memory slowly goes down over time, but not enough to run out) and then a few hours later I started getting brief Internet outages and finally no network activity at all in the middle of playing a network game on my iPad.

I couldn’t connect to Wifi after disconnecting, anything connected to Wifi or Ethernet couldn’t access the router, LAN or WAN. Everything was just dead. I power-cycled and everything was fine again. There was nothing in the logs. I never had issues prior to 384.5. I ran 384.4 for months without issue.

Any idea what’s going on?

I had an AC88U router appear to die last year. LED lights did not appear. After several attempts to get it working failed, I boxed it up so I could return it to the store. It sat inside the box over the weekend. Before leaving for the store on Monday, I decided to plug it in. Well, it came back online and has not given me an issue since then. Try turning it off, unplug it from the power source, and leave it in that state over night. Hopefully, it will be that simple.

 

[Morac]

I had an AC88U router appear to die last year. LED lights did not appear. After several attempts to get it working failed, I boxed it up so I could return it to the store. It sat inside the box over the weekend. Before leaving for the store on Monday, I decided to plug it in. Well, it came back online and has not given me an issue since then. Try turning it off, unplug it from the power source, and leave it in that state over night. Hopefully, it will be that simple.

I’m guessing you didn’t read my post and only read the title since the router hardware isn’t broken, it just locks up every 2 weeks.

 

[indark]

you mention about resetting, have you tried clear nvram?
it sometimes help those weird issue.

you can either use ssh or use the wps button to do it.

my preferred method is just power off the router for 30 sec, hold wps button while powering up.
wait till power led start blinking and release it.
wait till it finished reboot then reconfigure your setting.

while it's not necessary step for flash firmware, i pretty much do it everytime I encounter weird issue with the firmware upgrade ;p

 

[scjr]

Morac,

Don’t know what could be happening with your setup, but 384.6 is definitely worth trying. Even though it’s an alpha, it’s super solid and the reviews in thread have been generally good. Worth a try. Did you start from a clean slate with 384.5?

 

[Xentrk]

I’m guessing you didn’t read my post and only read the title since the router hardware isn’t broken, it just locks up every 2 weeks.

Nope, I read the post. Thanks for clarifying the issue. I have three AC88Us running 384.5 with no problem. Do you have USB drive attached? Did you run a disk check on the external drives? The only other thing I can think of is to run the top or htop command to see what is running and taking up memory.

 

[ARol]

One of my friend's RT-AC68U has this problem too,
the diagnostics is that exactly after 14 days uptime all reads and writes to jffs are stopped,
web interface shows 100% cpu on one core, "ps w" shows all processes reading/writing to jffs in D state,
like

 # ps w|grep " D "|grep -v grep
xxx admin     yyyy D   cp /tmp/syslog.log /tmp/syslog.log-1 /jffs/
xxx admin     yyyy D  TrafficAnalyzer -e
xxx admin     yyyy D  [ -f /jffs/configs/profile.add ]

then even ssh login not working because of check [ -f /jffs/configs/profile.add ] in D state,
ssh admin@router 'uptime' shows LA 10+,
df -h shows jffs is not full (9.6M usage),
ls /jffs not responding,
my crontab (cru -l) is empty,
no errors in dmesg or syslog.log,
reboot is only possible by power cycle...

First time it happens I noticed last lines from syslog.log are lost and strange error in log after reboot

kernel: JFFS2 notice: (145) check_node_data: wrong data CRC in data node at 0x02b74f68: read 0xcca1912c, calculated 0xad01fa8c.

Next time after 14 days I got 100% cpu again and I followed this post

• backup jffs scripts and configs folder
• check jffs off
• reboot
• check jffs on & check format jffs on
• reboot
• reboot
• restore scripts files from backup
• reboot

some last lines from log also lost, but no more JFFS2 errors after reboots, now waiting for 14 days again...
Any Ideas?
What special may happen in fw at 14 day uptime?

 

[ARol]

So, now I have this problem again and uptime is only 2 days..(( Now I'll try to reset, reflash, factory reset and reconfigure

 

[agilani]

So, now I have this problem again and uptime is only 2 days..(( Now I'll try to reset, reflash, factory reset and reconfigure

I had this problem and it ended up being radio interference on my equipment. make sure you manually pick 2.4ghz channel that has the least conflict or use 5ghz. also check and see where you are writing traffic logs to. and upgrade to the latest code base.

 

[ARol]

I had this problem and it ended up being radio interference on my equipment. make sure you manually pick 2.4ghz channel that has the least conflict or use 5ghz. also check and see where you are writing traffic logs to. and upgrade to the latest code base.

Hi,
so, I have upgraded to latest 384.6, did a factory reset (twice, before and after flashing), also did 'nvram unset jffs2_size' and got this problem again after 14 days and some hours uptime... I have no idea why this problem related to radio and why router's flash lives only 14 days, router has both 2.4 and 5ghz bands enabled, with guest networks on each, 2.4 channel is manually set to channel 11 while other APs in the building set to channels 1 and 6. My TrafficAnalyzer.db is on jffs partition, but it's only 10Mb size..

For now I've enabled reboot scheduler weekly as a workaround, I think this is a hardware bug with flash:-(

 

[ARol]

I also tried

dd if=/dev/random of=/jffs/tmpfile bs=1M count=40 && \
rm -f /jffs/tmpfile

with no errors

And one more confusing thing is that df and du shows different values (but I've seen this on working routers too)

# df -h /jffs/
Filesystem                Size      Used Available Use% Mounted on
/dev/mtdblock4           62.8M      6.3M     56.5M  10% /jffs
# du -hs /jffs/
11.9M    /jffs/

still have no solution...

 

[ARol]

So finally I resolved problem with dying jffs by mounting usb flash used for optware over it

# cp -a  /jffs/scripts/post-mount  /jffs/scripts/post-mount.bak
# cat <<EOF > /jffs/scripts/post-mount
#!/bin/sh

if [ "$1" = "/tmp/mnt/opt" ] ; then
  ln -nsf $1/entware /tmp/opt

  #try to free jffs usage
  mount -o bind /tmp/mnt/opt/jffs /jffs

fi
EOF
# chmod +x /jffs/scripts/post-mount
# cp -a /jffs /tmp/mnt/opt/
# reboot

maybe later write more scripts to sync back from usb to jffs