Machine Learning container not downloading new models
I know this worked in the past, it may be related to the most recent 1.133.0 update, or some other issue. My setup is running as non-root user, and everything was working perfectly before. Now, when I try to change to a different CLIP model, the machine learning container start downloading the model and creates some files and folders, but at some point, it fails with
RuntimeError: Data processing error: I/O error: Permission denied (os error 13)
The file permissions on the model-cache folder appear correct, the 1000 user seems to have full access to everything, and the folders and files created when the download of the model starts are owned by that user as well....not restarting after reboot
Hello
I also posted this question on the discussion page as I'm not exactly sure where the correct place is (https://github.com/immich-app/immich/discussions/18533).
But basically my problem is this:
When I set any docker deploy setting (see below for my docker compose setup), the containers which have the deploy settings set won't restart after a system reboot or after a restart of the docker service. If I remove the deploy section but keep everything else, it works as expected. It seems like the deploy section does not consider my restart policy. The deploy setting itself works as I can see my graphics card inside the containers.
```yml...
Error: The pgvecto.rs extension is not available inthis Postgres instance.
Running postgresql:17 on another machine
Installed vchord and required vector extension.
Using DB_URL='postgresql://immich:<password redacted>@10.5.0.111:5432/immich_db'...
Need help upgrading to 1.133 in CosmOS
Im not able to get immich up and working. Since cosmos uses json instead of yml, I'll leave them down below. The problem comes when I try to remove the command section from the immich-database container, because the it wont start and then the immich-server will keep looking for it.
Changing the database location
I want to change the database location from the pgdata docker volume to explicitly live in a folder on my server. I first backed up the pgdata volume to an external hard drive to avoid mishaps. I then copied the contents of the pgdata volume to a folder where I want the database to be stored. I then updated the DB_DATA_LOCATION in the .env file to the new path.
I notived however, that in the docker compose file it still says
```
volumes:...
1.133.0 Change location button does nothing
Version: v1.133.0
Browser: Ubuntu snap 136.0.2 (64-bit)
When I mark some photos and then click at the top right corner (Menu) and "Change location" it does nothing.
On the network tab I can see it loads the tiles from the tileserver but I get nothing shown on my screen.
Is that a known bug or am I doing something wrong?...
In 1.133.0 I can no longer authenticate with Authelia
Both sides have the
token_endpoint_auth_method: 'client_secret_post' set. But when Authelia returns to immich it says failed to finish auth with the following error:
```
[Nest] 17 - 05/23/2025, 7:48:40 AM ERROR [Api:OAuthRepository~dweqkoeg] OAuth login failed: server responded with an error in the response body
[Nest] 17 - 05/23/2025, 7:48:40 AM ERROR [Api:OAuthRepository~dweqkoeg] ResponseBodyError: server responded with an error in the response body
at checkOAuthBodyError (file:///usr/src/app/node_modules/oauth4webapi/build/index.js:865:19)...Setting up pin results in 500 error
No way to reset it either. Can anyone help me please? there are no photos behind it right now.
Machine Learning URL?
After updating to version 1.133.0, machine learning was disabled, i then enabled it, and it stopped working, i have tried the following urls
(10.0.0.63 is local ip, port 6379 is redis, which google said it helps)
```
http://10.0.0.63:3003...
Android app thinks that nothing is backed up.
After updating both app and server, the android app in the sync tab now believes that none of my files are backed up, wanting me to backup over 6000 files all over again.
The playstore didn't recognize the update so I uninstalled and reinstalled the app to update, losing my settings, perhaps this is why?...
New external web domain dependency in v1.133.0?
I'm not sure where to ask this, so I'm starting here.
It looks like every web page on the latest version now has a <script src> link to
https://www.gstatic.com/cv/js/sender/v1/cast_sender.js?loadCastFramework=1.
I don't think this dependency existed in the past, and other than this, I don't think there are any external web dependencies for Immich....Failed to activate VectorChord extension
Encountered an issue after upgrading to V1.133.0 - immich_server container continuously restarts. I checked logs with:
and got the following error from the log:
```
Failed to activate VectorChord extension....
docker log immich_server
docker log immich_server
Videos have red exclamation mark
After I put some videos that were previously Archived and Favorited into the Locked Folder and back into the main timeline, I can no longer watch them. It seems like they are frozen and when I hover over the play sign with the mouse, I only see the red exclamation mark. How can I repair the files? I don't see anything in the logs when I try to watch the video.
Thanks for any help
```...
Issue with Immich Upgrade - Incompatable database?
I followed the releases guide in upgrading but i'm stuck with seemingly an incompatable database?
Error:
Server is also in a reboot loop...
The data directory was initialized by PostgreSQL version 14, which is not compatible with this version 16.8 (Debian 16.8-1.pgdg120+1).
The data directory was initialized by PostgreSQL version 14, which is not compatible with this version 16.8 (Debian 16.8-1.pgdg120+1).
"Error loading image" when skipping through photos too fast?
When I open a picture and then skip through the images rather fast with the arrow keys, I get "Error loading image" quite often. Moving one photo back and then forward again then loads the picture successfully. Going through the images slowly works fine, all thumbnails in the timeline load correctly.
Mobile app also does not show this issue.
Does anybody experience this as well? Is there some fix for this?...
0.133.0 tries to DROP VectorChord extension instead of using it
I have external Postgres instance. Followed upgrade to VectorChord procedure. During first startup I got:
LOG [Api:DatabaseRepository] Dropping VectorChord extension
Why is it behaving like this instead of migrating database to new extension?
This is Kubernetes self-managed installation but I don't think it makes any difference while using external database....
Set.prototype.difference is not a function
Hi all, I just upgraded to v1.133.0 and can no longer use the web interface. It loads, but gives the error "Failed to load assets". The mobile app works fine.
I see there is another support post for this same issue, where the user is on an old version of Firefox which doesn't support Set.prototype.difference.
I am on Chrome 109, on a device that unfortunately cannot be upgraded to a newer version of Chrome. ...
Failed to activate VectorChord extension. Please ensure the Postgres instance has VectorChord in
Hello i migrate my docker compose .yml to the new image.
When immich_server start i have this error :
Nest] 7 - 05/22/2025, 8:12:07 PM FATAL [Microservices:DatabaseService] Failed to activate VectorChord extension.
Please ensure the Postgres instance has VectorChord installed.
...
1.133.0 update help
Hi everyone,
I’m a beginner user and a bit inexperienced with Immich. I’ve carefully read the changelog for version 1.133.0, but I’m a little worried about causing damage during the upgrade. Could someone help me for a few minutes so I don’t mess everything up? XD
Right now I’m using Immich in a very simple way: I installed it on Ubuntu Desktop (accessed via SSH) and configured an external drive on my NAS as the storage location. I haven’t changed any other settings, so I’m not sure if the upgrade instructions are meant for “basic” setups like mine or only for more complex configurations....