#Error: type "vendorindexinfo" does not exist

1 messages · Page 1 of 1 (latest)

next parrot
#

I'm currently working to upgrade my immich instance to 1.95.0, but I'm struggling to get my pgvecto.rs instance to upgrade to v0.2.0. Here is my setup:

Currently running Cloudnative pg in kubernetes. I swapped out ghcr.io/bo0tzz/cnpgvecto.rs:14-v0.1.10 for ghcr.io/tensorchord/cloudnative-pgvecto.rs:14.11-v0.2.0. I simply changed the image and commited the changes and CNPG rolled out the changes to my cluster. I then followed the migration steps in the 1.95.0 release notes. When I run ALTER EXTENSION vectors UPDATE TO '0.2.0';, I get an error that says ERROR: type "vectorindexinfo" does not exist. Then running SELECT pgvectors_upgrade(); fails.

Postgresql config for kubernetes: https://github.com/ModestTG/jace-cluster/blob/main/kubernetes/apps/database/cloudnative-pg/cluster/cluster2.yaml

I haven't restarted the PG cluster yet as it requires me to bootstrap from backup. Do I need to do that or potentially stream the database to a new cluster with the new image? Any help is appreciated. Thanks!

Edit: the user that access the immich database is not a super user

tawdry bloomBOT
#

:wave: Hey @next parrot,

Thanks for reaching out to us. Please follow the recommended actions below; this will help us be more effective in our support effort and leave more time for building Immich immich.

References

Checklist

  1. :ballot_box_with_check: I have verified I'm on the latest release(note that mobile app releases may take some time).
  2. :ballot_box_with_check: I have read applicable release notes.
  3. :ballot_box_with_check: I have reviewed the FAQs for known issues.
  4. :ballot_box_with_check: I have reviewed Github for known issues.
  5. :ballot_box_with_check: I have tried accessing Immich via local ip (without a custom reverse proxy).
  6. :ballot_box_with_check: I have uploaded the relevant logs, docker compose, and .env files using the buttons below or the /upload command.
  7. :ballot_box_with_check: I have tried an incognito window, disabled extensions, cleared mobile app cache, logged out and back in, different browsers, etc. as applicable

(an item can be marked as "complete" by reacting with the appropriate number)

If this ticket can be closed you can use the /close command, and re-open it later if needed.

tawdry bloomBOT
honest kayak
#

I think @ornate leaf ran into something similar

gilded rock
#

But I don't know if this necessarily means it'll be as problematic for you. You can try the things in that thread and see where it goes

next parrot
ornate leaf
#

I'm not sure if there is an upgrade path from 0.1.10 - 0.2.0

next parrot
#

Is it not possible to stream the data into a new CNPG instance with 0.2.0 installed? Or do you have to alter the tables still to make it work?

next parrot
#

This might be a dumb question, but what would I lose if I just scrapped my DB and started over. All of the picture originals are there. It would just be all of the ancillary stuff (users, faces, albums, etc.). Obviously I'd like to avoid this but at least for myself I could rebuild all of that, even though it would be very annoying.

#

When you assign a date from the Immich UI, does that actually update the exifdata on the image or just in the DB?

ornate leaf
#

I don't think you would need tos tream into a new instance for that

ornate leaf
next parrot
ornate leaf
#

Yes, the steps from the release notes include creating that schema

next parrot
#

Right. I created the schema but they aren't listed for the tables in the database.

#

So the schema is listed there but it's not in the "schema" for the actual tables. Not sure if that's an issue because it looks like my instance was upgraded and potentially working maybe kind of sort of nobody breathe?

ornate leaf
#

I -think- vecto.rs will create a view in the vectors schema that would show up there? But @gilded rock will know for sure

#

If everything is working including CLIP/faces then I would just call it a day :P

next parrot
next parrot
ornate leaf
#

For CLIP just do a context search and see if things explode

#

For faces the easiest is probably to just upload a new image with a face in it

gilded rock
#

The vectors schema is just for the extension. immich tables will still be public

next parrot
#

So I just uploaded a selfie and got these errors. These seem consistent with the errors I saw when trying to reindex the database

gilded rock
#

You probably ran DROP EXTENSION vectors CASCADE. I wish Postgres didn’t suggest that…

next parrot
#

I did 😦

#

Any remedy for that or no?

gilded rock
next parrot
gilded rock
#

Haha yeah you’re fine

#

The model you’re using is shown in the admin settings, but if you didn’t touch it then no need to worry about it

next parrot
#

I'm running into the same issue that Foritus on github ran into where I'm getting a permission issue when I run CREATE INDEX IF NOT EXISTS clip_index ON smart_search USING hnsw (embedding vector_cosine_ops) WITH (ef_construction = 300, m = 16);. I tried ALTER EXTENSION vectors OWNER TO pg_database_owner but that didn't work. I'm not sure what piece of software he was using to manage his postgres instance.

https://github.com/immich-app/immich/discussions/7381#discussioncomment-8573375

#

I got that fixed. Needed to change the schema not the extension. Once I got that fixed I reran the SQL and got this.

#

Do I now need to run the upgrade commands from that link do you think @gilded rock?

#

I get no rows returned from the pgvectors upgrade instructions. Not sure how to delete all index files. Maybe they're already deleted and everything is empty and I'm good to go?

next parrot
#

So it's telling me both that I don't have any index files, but also that the extension is upgraded an all index files are outdated and need to be deleted.

gilded rock
#

Try running SELECT pgvectors_upgrade(); and restarting

next parrot
#

I brought down my cluster and brought it back up. I got the same results.

#

I imagine clip_index and face_index need to be in here right?

next parrot
#

Well I feel like I'm getting somewhere. I tried restarting postgresql (brought down all 3 instances of CNPG and brought back up). Still seeing this: