On 2025-07-22 my apps started failing. Failing in the sense of them not running / not starting up. At first it was just 1 out of 41 (at least I only noticed it for 1). I tried restarting that app and it failed to do so. Then I started troubleshooting different things, starting with just rebooting the server, and removing old docker images. Ultimately (i.e. now) I only have 7 of 41 running.
Trying to start an app immediately gives me an immediate fail to start error and tells me to look at the app_lifecycle. log.
Any app I try to start gives me the following (I am just copypasting a single example, draw. io):
[2025/07/23 07:36:02] (ERROR) app_lifecycle.compose_action():56 - Failed ‘up’ action for ‘drawio’ app: Network ix-drawio_default Creating\n Network ix-drawio_default Error\nfailed to create network ix-drawio_default: Error response from daemon: cannot create network ad441acb51bbfd7745d172a0d8694583e9d8c96edd345d66d84d37346dd2a2f8 (br-ad441acb51bb): conflicts with network 3bf8116c15f1b76849aad17d2a591c37c4482a0ec8eb11367e7c914894feeb20 (br-3bf8116c15f1): networks have overlapping IPv6\n
I am not sure if there is a causation, but I will share a timeline of recent activities regardless:
2025-07-16: I update Truenas Scale (CE) from 24.10.2.2 Electric Eel to 24.10.2.3, and shortly after to 25.04.1 Fangtooth
2025-07-22: I update my router/switch Fritzbox 7590 to Fritz!OS 8.20 at 09:32 in the morning.
Since the error messages point to something ipv6 related, and because I started noticing more and more failures within 24h of the update I thought it worth mentioning the firmware update on the router.
Any ideas on how to fix this? The apps have been running rock solid since upgrading away from Kubernetes. I don’t even actively use ipv6, so if I could just turn that off entirely, that could be an easy fix. My setup is exclusively used locally, on the home LAN.
If you don’t rely on any ipv6 networking i’d simply switch it off in the fritzbox and truenas and see if the apps start again when no ipv6 is available.
I should have added it to the OP, but I had already tried changing setting on the router in regards to ipv6 (a little bit). But, tbf, I find the settings in Fritz always to be a bit vague. So, I just tried again, with a heavier hand. Afai can tell I have deactivated any active responsibilities about managing ipv6 in Fritz.
But it does not seem to change anything. I was hoping I could turn off ipv6 in Truenas CE. Perhaps just for the apps? But I don’t see anything obvious in Truenas settings either. Is there something I can do in the app settings and/or docker compose?
My guess is that whatever entity in my network is responsible for handing out ipv6 addresses is doing a poor job.
Does the lifecycle log ordinarily clear all issues once they are successfully completed? Would those extraction states still being there indicate they are stuck?
In Apps > Configurations > Settings there are Address Pools.
I had never touched them since starting out with this server and Truenas Scale 2-3 years ago.
There were two entries there. A second one with an IPv6 Base. I removed it. Now apps are starting up again!
It was never an issue for over 2 years, until it was.
For the small chance someone runs into something similar, there is a possibility that it was the unique combination of the above setting with the current version of Nginx Proxy Manager (2025-07-24) that messed things up.
I can’t conclusively figure it out, but all signs point in that direction.
Another few days later, I have a suspicion that another thing that added to the issue was a setting in Adguard Home. After thinking I had fixed my issues, I noticed that my internet was almost always really slow. I often had to reload my browser for it to actually load a page.
I looked at Adguard Home more closely and noticed upstream response times of 900ms and even much longer. Something must be up with DNS stuff.
By default “use private reverse DNS resolvers” is activated in Adguard Home. It seems like this created some loop where all/most internet traffic would look within my network first, specifically Home Assistant OS.
Again, this setup had worked perfectly fine for 3 years or so. Suddenly it became an issue that spread and probably created odd issue all over the place.
