hAP ax3 – Netwatch causes high CPU usage after several days (ROS 7.21.1)

RouterOS
1

Hello,

I have an issue with hAP ax3 running RouterOS 7.21.1.

After several days of uptime, CPU usage suddenly becomes very high. The load is mostly caused by the netwatch process.

I only have two netwatch entries configured:

/tool netwatch
add check-certificate=yes disabled=no down-script="" host=1.1.1.1 http-codes=301 name=1.1.1.1 test-script="" timeout=5s type=https-get up-script=""
add check-certificate=yes disabled=no down-script="" host=8.8.8.8 http-codes=302 name=8.8.8.8 test-script="" timeout=5s type=https-get up-script=""

Important observations:

  • Disabling the netwatch entries does NOT help.

  • Removing the netwatch entries completely reduces CPU load after about 10–20 minutes.

  • Rebooting the router immediately fixes the issue.

  • After removing the entries, waiting until CPU normalizes, and then recreating them — everything works fine again for several days, until the problem repeats.

Attached is a CPU graph from the previous occurrence of the issue (you can see long periods of ~90–100% CPU usage).

Has anyone experienced similar behavior with netwatch (https-get + check-certificate=yes) on ROS 7.21.1?

Could this be a memory leak or stuck HTTPS session inside netwatch?

Thank you.

изображение
изображение1483×396 40.8 KB

2

изображение
изображение908×590 39.9 KB

3

Too many unfinished and not timeouted netwatch tasks? Memory exhausted with these tasks?

4

изображение
изображение1072×539 89.1 KB

5

try to increase interval from 10 second (default) to something bigger

6

I removed the netwatch entries and waited about 10 minutes — CPU usage returned to normal.
Then I added new netwatch rules with an interval of 1 minute.
Now I’m monitoring the situation and waiting to see if the issue reappears.