I had it once a few days back but I was juggling different versions, so not really sure which one was that. In general my impression is -ct driver is unstable and crashes frequently.
I agree that -ct is not ready for prime time. My problems happen on "old" though.
I am new to openwrt and my experience so far isn't bad. Still from time to time some process (usually nlbwmon) is going wild and causing high CPU utilization. Therefore I decided to put some simple monitoring and alerting in place. I am putting it here since it has been tested with R7800 only but should be pretty universal.
- Sendmail first (configured for outlook.com)
opkg install ssmtp
/etc/ssmtp/ssmtp.conf
root=myemail@outlook.com
mailhub=smtp-mail.outlook.com:587
rewriteDomain=outlook.com
UseStartTLS=YES
AuthUser=myemail@outlook.com
AuthPass=myOutlookApplicationPassword
/etc/ssmtp/revaliases
root:myemail@outlook.com:smtp-mail.outlook.com:587
- And last but not least following line to be added to /etc/rc.local before exit 0
t=$(while true; do T=$(sleep 15m; if [ `cat /sys/devices/virtual/thermal/thermal_zone9/temp` -ge 60000 ] && [ `sleep 15m; cat /sys/devices/virtual/thermal/thermal_zone9/temp` -ge 60000 ]; then echo -e "Subject: RouterName Thermal/High CPU Alert\n\n"`cat /sys/devices/virtual/thermal/thermal_zone9/temp`"\n"`uptime`"\n"`top -b -n 3`"\n" | sendmail mymail@gmail.com; sleep 3h; fi); done) &
First sleep is to ignore higher CPU load at the boot-up and governs also how often temperature is being checked. Second one to send alerts only if higher temperature is observed within 15 minutes time frame. Third one to send alerts not more often than every 3 hours.
I found it easier to monitor temperature however script can be modified to check actual CPU utilization instead.
Mail being sent contains some diagnostic info (result of uptime overlaps with top but is easier to be read).
Threshold temperature set to 60 Centigrade.
1 Like
Same issue here with 18.06.2 and nlbwmon.
Version with the uptime is simpler. Everything is the same except of entry in the /etc/rc.local:
(while true; do (sleep 15m; ([ `uptime | awk '{print (int($8*100))}'` -ge 300 ] && (echo -e "Subject: RouterName Load Alert\n\n"`cat /sys/devices/virtual/thermal/thermal_zone9/temp`"\n"`uptime`"\n"`top -b -n 3`"\n" | sendmail mymail@gmail.com); sleep 3h)); done) &
Treshold for 15 minutes load average is set to 3.00.
do you only notice this while viewing load average from luci? If so, have you seen this?
an attempt to set option max_requests 1 was made but was reverted (see here) as concurrent requests are apparently needed. I tried setting max_requests to 1 and my symptoms went away without other issues but I've sent set it back to 3 as I don't use luci much...
HTH
Might be however could just be coincidence as well.
Once it starts the only remedy is to restart nlbwmon. If so the issue is in implementation and not simply with number of concurrent requests. Otherwise it would be straight forward to reproduce but it's not.
Since this issue is not specific to the community build that this topic is about, it would be best to open a new topic for this issue.
owrt1907-r10199-f96fbf0328-20190612
First build from the stable openwrt-19.07 branch.
Currently there is not much difference to master, but the 19.07 will soon start to deviate from master, as development goes on in master while 19.07 remains the stable release branch.
Based on the branch name, the official 19.07.0 release is expected for July (but can get postponed to August).
I will soon start to phase out the 18.06 builds.
NOTE: there is no 19.07 buildbot yet, so there are no add-on packages for 19.07. If you need to install something, you could try downloading and installing master snapshot packages, and possibly force the install and hoe that the package works...
1 Like
Thanks for the new stable build @hnyman! I have been successful in recreating the build environment and able to create images, but I've searched through the forum. Although I haven't found if it was possible to modify the script to be able to build a 18.06.2 stable build with your patches if someone can direct me to what I need to modify in the script if it's possible sorry still new to all of this not sure if I should have posted here or created a new thread. Thanks everyone!
Easiest would be to simply download the scripts from my newest 18.06 build and to create the build env with them. 18.06 requires different patches than master, so you need those anyway.
And note that likely you should build from the 18.06 head to get the newest fixes, instead of the half-year old 18.06.2 release. If you really want the old 18.06.2, read advice from this thread
I'll stick with the scripts how you said they really do help clear things up and stay up to date with the newest fixes. Thank you!
hello,
do i need to update your patches? (i'm using patches i thing a couple of months old)
should i also re-create the build environment?
or can i simply go on building in the same way (with my menuconfig and packages' modifications)?
thanks a lot
Most likely, yes, as the underlying sources have changed to some extent.
If there has been change in sources, the patched files may need to be changed / edited /modified as necessary, if you git pull updated sources from upstream.
Quite normal git work...
Easiest may be to re-create the whole thing.
Is 19.07 stable going to be ath10k-ct only?
Not sure yet if I will build both ath10k versions for it.
-ct has lately worked ok for me.
After almost a year I updated your build from 84xx to the new 19.07 10199. Great job so far with the old build!
Unfortunately I cannot connect more than 32 wifi devices:
"refusing to associate station: too many connected already (32)"
Anyone any idea how to increase the number?
Using now the 10192 ath10k and it works again 
The latest version of ath10k-ct is way more stable, but I still experience periodic disconnects and need to re-connect devices once a day or so.