Selfhosted version won't connect to network endpoints after I tried to force API 1.41

I am very new to posting here. I did not find any topic related to mine but i must confess I am a bit overwhelmed with all the information available here and may very well have overlooked something. I am not a native english speaker.

I self host bitwarden on a bookworm VPS to manage my passwords and today, my instance won’t start with messages saying `Error response from daemon: Container cannot be connected to network endpoints: docker_default, docker_public`.

Here is what I did to end up in this situation.

Every now and then, i upgrade my system and instance with the following commands:

apt upgrade
./bitwarden.sh updateself
./bitwarden.sh update

The system upgrade upgraded the following docker related packages:

  • docker-buildx-plugin → 0.31.1-1~debian.12~bookworm
  • docker-compose-plugin → 5.1.0-1~debian.12~bookworm
  • docker.io → 20.10.24+dfsg1-1+deb12u1+b3

The upgradeself command updated from 2025.12.1 to 2026.2.0

And the latter update command said: Error response from daemon: client version 1.53 is too new. Maximum supported API version is 1.41.

It is possible that the whole thing have been ending up with this message for quite a while now and that just today I noticed it. I don’t know. Anyway, at this stage, I believe that my bitwarden instance has not been updated but is still running fine on version 2025.12.1 (unless the upgrade was already failing before in which case I have no idea what version of bitwarden I was running and have no way to find out as far as I know).

I know that I can enforce the 1.41 API by setting DOCKER_API_VERSION=1.41 so I did that and ran ./bitwarden.sh update again. New images were pulled, as usual, but when starting, the following error occurred:
```
[+] pull 90/90
:check_mark: Image ghcr.io/bitwarden/events:2026.2.0 Pulled 28.1s
:check_mark: Image ghcr.io/bitwarden/attachments:2026.2.0 Pulled 25.3s
:check_mark: Image ghcr.io/bitwarden/notifications:2026.2.0 Pulled 43.2s
:check_mark: Image ghcr.io/bitwarden/api:2026.2.0 Pulled 30.2s
:check_mark: Image ghcr.io/bitwarden/icons:2026.2.0 Pulled 38.0s
:check_mark: Image ghcr.io/bitwarden/sso:2026.2.0 Pulled 38.3s
:check_mark: Image ghcr.io/bitwarden/mssql:2026.2.0 Pulled 4.7ss
:check_mark: Image ghcr.io/bitwarden/web:2026.2.0 Pulled 46.2s
:check_mark: Image ghcr.io/bitwarden/admin:2026.2.0 Pulled 41.5s
:check_mark: Image ghcr.io/bitwarden/identity:2026.2.0 Pulled 43.7s
:check_mark: Image ghcr.io/bitwarden/nginx:2026.2.0 Pulled 46.1s
[+] up 5/11
:check_mark: Network… Created 0.3s
:check_mark: Network… Created 0.7s
⠋ Contain… Creating 0.0s
✘ Contain… Error response from daemon: Container cannot be connected to network endpoints: docker_default, docker_public 0.0s
⠋ Contain… Creating 0.0s
✘ Contain… Error response from daemon: Container cannot be connected to network endpoints: docker_public, docker_default 0.0s
⠋ Contain… Creating 0.0s
⠋ Contain… Creating 0.0s
⠋ Contain… Creating 0.0s
⠋ Contain… Creating 0.0s
✘ Contain… Error response from daemon: Container cannot be connected to network endpoints: docker_default, docker_public 0.0s
```

At this point, I figured I should abort my upgrading process and take some time to think about what is going on. But I am unable to downgrade to 2025.12.1 (ie. download bitwarden.sh version 2025.12.1 from Release Version 2025.12.1 · bitwarden/self-host · GitHub and use it to call update). I have the exact same symptoms. I tried to go back in time as far as 2025.6.0 with no difference.

So here I am asking for help. My priority is to have my instance running again. I will troubleshoot why the upgrade fails later.

Thanks for your help.

I have no idea how to solve my problem and downgrade, but upgrading docker with up to date ce versions instead of native debian apt packages solved the problem (my instance is version 2026.2.0 and up and running).
To do so, I stopped the docker service, removed all containers and networks, then followed these instructions : Debian | Docker Docs.

2 Likes