I have pihole running in a docker container with a lxc container as the host. All vm's and lxc's are hosted on a simple proxmox server with local storage - no cluster and/or zfs involved.
There is weekly update process (friday/saturday night) assuring that everything is running at the latest stable versions.
All vm's and lxc's are backed-up twice a week with pve in stop/start mode utilizing seperate, plain lvm storage. Meaning pve is setup in a way that snapshots are not supported.
After the backup of the lxc with pihole, it looks like the reverse lookups are not always working - in particularly the ones that involve tailscale nodes - sometimes also others from devices in local subnets. Meaning the dashboard shows ip's versus dns names.
After restarting the pihole container with docker restart pihole everything is back to normal.
Does this behavior ring any bells? Any idea what could be happening?
Anyone an idea?
I also noticed the setting misc.delay_startup:
To prevent delayed startup while the system is already running and FTL is restarted, the delay only takes place within the first 180 seconds (hard-coded) after booting.
What is used as a reference for a "system already running" and "after booting"?
I think simply what it says :
- System already running = Uptime over 180 seconds which is 3 minutes.
- After booting = After the login screen shows up or the last service has been started by the system during the boot process.
I can't imagine it being more complex than that ?!
You are right - it is not more complex then what it says.
But my question is not about what it says - it is about what is the decision based on?
What variable is used to determine if a system is already running? Or if a system is rebooted?
Rationale: since an lxc container is relying on the pve-kernel I can imagine a reboot may not being noticed unless pve itself is rebooted. Hence the question.