I tried to use containers under the Mikrotik router os. hAP ax3 - ROS version: 7.22.1.
The container, pihole is working well, I pulled and installed it to en external USB pendrive. This is the recommended option to use a container to save the internal flash drive lifetime.
But the write operations number are much higher as earlier. Before the container installation the sector write counter is about 45-50 sector/h, after the container is starts and works, this counter is 800-900 sector/h.
That's quite a big increase.
All log operation is set up to log to the memory.
Container mounts:
/etc/pihole --> Munted to usb1/containers/pihole/pihole
/var/log/pihole --> Munted to usb1/containers/pihole/log
/etc/dnsmasq.d --> Munted to usb1/containers/pihole/dnsmasq.d
Shared memory size: 64 - 128 MiB (It seems the 64 Mbyte is enough.)
ram-high: none
Is this sector/h writing number is a normal operation, or did I missed/forget something (and does the container saves any data to the internal flash)?
How exactly? Did you set in Pi-hole config all for log files /dev/null (there is an issue if you do that in recent Pi-hole versions and Pi-hole won’t start if FTL log file points to /dev/null) or did you mount /var/log/pihole dir as tmpfs (as I see from above post it is not)?
Anyway if you properly set container root dir. path into USB drive (not shown above) it shouldn't affect internal drive.
Sorry, I didn't make it clear that the system logging is written to memory, so the system logging definitely doesn't write to the internal flash drive. It seems the log of the container is written to the memory too.
The logs of the Pi-hole are on the external drive.
Container - Mounts: Src: usb1/containers/pihole/log Dst: /var/log/pihole
(pihole.log, FTL.log, webserver.log,…)
Root dir: /usb1/container_images/pihole
Only one thing is not clear to me, the Workdir:
”workdir (string; Default: ) the working directory for cmd entrypoint”
Value: /
If is configured as stated then it shouldn't write to internal flash. I’m running Pi-hole also in container and don’t have such issues, container root dir is on USB drive along with /etc/pihole mount for persisting configuration after upgrade, /etc/dnsmasq.d mount is not needed for Pi-hole v6 because Dnsmasq config is also inside /etc/pihole dir.
If you are sure that Pi-hole container is causing this issue (multiple stop/start container monitoring) IDK what can be the issue unless something is corrupted internally in ROS containers which is causing this (check container debug logs if something can be found) and in that case only Netinstall helps.
Yes, the container is working with healthchecks, so it can be the reason of this issue.
I will check the disable option, or I will modify the intervals of healthchecks.
I can wait for final resolution, because I just install Pi-hole container for testing purpose.
