Open lmgithu opened 3 months ago
some additional observations:
If i start another migration to migrate thumbnails for assets and faces to the latest folder structure, that job always runs on ALL my assets (100K+)... each time i see the huge number of assets that go down to zero but it seems it is not getting processed in reality as each time when i start the job it shows same high number.
Can you help double-check the files in the older format yyyy-mm-dd
to see if the path corresponds to the file's path that is shown in the UI?
I think the issue is now even bigger, I tried to check a file in that folder and one of them is there on the file system but NOT on the UI.. I don't see these images anymore :( In another case I do see as you do, it is on the UI as well and it still refers to the old folder (attached screenshot below)
If you need, I can share the logs and/or the file itself that it did not process. Let me know how I could help.
NOT on the UI.
Can you use the search for file name to find that file to see if it returns in the search result. Please attach the screenshot of the file on the file system as well as the search result if you can
Also logs, on the server when you run the migration would be helpful, you can increase the log level to debug in the Server settings in the Administration page
Will do and report back in an hour. How can I download/extract the log file?
From the terminal, run docker logs immich_server
or docker logs immich_microservices
hi Alex,
Here are the results: searching for the image name on UI: no result. I can't take a screenshot as there is no results, it just shows my photo feed (I confirm it works for other image names). Screenshot of file system/location of the image: attached.
Log:
immich_server_1 | [Nest] 6 - 03/22/2024, 6:25:05 PM DEBUG [JobService] Handling command: queue=storageTemplateMigration,force=false
immich_redis_1 | 15:C 22 Mar 2024 18:05:08.034 * RDB: 0 MB of memory used by copy-on-write
immich_redis_1 | 1:M 22 Mar 2024 18:05:08.107 * Background saving terminated with success
immich_redis_1 | 1:M 22 Mar 2024 18:10:09.036 * 100 changes in 300 seconds. Saving...
immich_redis_1 | 1:M 22 Mar 2024 18:10:09.037 * Background saving started by pid 16
immich_redis_1 | 16:C 22 Mar 2024 18:10:09.161 * DB saved on disk
immich_redis_1 | 16:C 22 Mar 2024 18:10:09.162 * RDB: 0 MB of memory used by copy-on-write
immich_redis_1 | 1:M 22 Mar 2024 18:10:09.239 * Background saving terminated with success
immich_redis_1 | 1:M 22 Mar 2024 18:15:10.025 * 100 changes in 300 seconds. Saving...
immich_redis_1 | 1:M 22 Mar 2024 18:15:10.025 * Background saving started by pid 17
immich_redis_1 | 17:C 22 Mar 2024 18:15:10.052 * DB saved on disk
immich_redis_1 | 17:C 22 Mar 2024 18:15:10.052 * RDB: 0 MB of memory used by copy-on-write
immich_redis_1 | 1:M 22 Mar 2024 18:15:10.126 * Background saving terminated with success
immich_redis_1 | 1:M 22 Mar 2024 18:20:11.071 * 100 changes in 300 seconds. Saving...
immich_redis_1 | 1:M 22 Mar 2024 18:20:11.072 * Background saving started by pid 18
immich_redis_1 | 18:C 22 Mar 2024 18:20:11.098 * DB saved on disk
immich_redis_1 | 18:C 22 Mar 2024 18:20:11.098 * RDB: 0 MB of memory used by copy-on-write
immich_redis_1 | 1:M 22 Mar 2024 18:20:11.172 * Background saving terminated with success
immich_redis_1 | 1:M 22 Mar 2024 18:25:12.096 * 100 changes in 300 seconds. Saving...
immich_redis_1 | 1:M 22 Mar 2024 18:25:12.096 * Background saving started by pid 19
immich_redis_1 | 19:C 22 Mar 2024 18:25:12.120 * DB saved on disk
immich_redis_1 | 19:C 22 Mar 2024 18:25:12.120 * RDB: 0 MB of memory used by copy-on-write
immich_redis_1 | 1:M 22 Mar 2024 18:25:12.196 * Background saving terminated with success
immich_machine-learning_1 | [03/22/24 18:25:17] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:25:27] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:25:37] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:25:47] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:25:57] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:26:07] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:26:17] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:26:27] DEBUG Checking for inactivity...
immich_machine-learning_1 | [03/22/24 18:26:37] DEBUG Checking for inactivity...
Just to make sure, can you drag-and-drop that file on the web to see if it successfully upload or it will say duplicated
I tried, it said "skipped 1 duplicated asset"
So it should be on the server, could it be in the archive view somehow?
No, it is not there in the archive :(
Can you confirm the visual of the image on that date,not just the file name?
Yes I can, I can't see this image visually on the UI, despite it is on the file system.
hmm strange, can you show me the search query you put in to search for the file name?
here you are:
this works for any other filename btw and I can see the results
Try 5841? and did you tick on the archive button at the very bottom of the page?
Do you know why certain folders/files were ignored? this is how my folders look like:
Interestingly, until 2010 every folder looks OK... after that, it is fully random and mixed.
I am not sure, we will have to look at this more in-depth. Thanks for the report
Could I somehow force process folders again?
or at least "delete" these files somehow so i could reupload them?
Can you check the Trash page to see if it is in it?
WIN, it was in the Trash! Sorry, this was such a noob issue. Now we "just" have the folder structure issue that it is fully random, I think there are no missing images. Do you have any guess what might be the issue and how to fix it?
Do you have any guess what might be the issue and how to fix it?
The issue is that we don't move files that are already in the trash :P So it is working as expect
no, it was a SINGLE image... not all of them.
99% of the files that were not moved are NOT in the Trash
Then we will need to circle back and look at this!
Thank you! Is there any way to "force" the migration again?
e.g. if i switch back to yyyy-mm-dd again and switch again to yyyy-mm afterwards? or via CLI?
Besides clicking on the template migration job, nothing else can be done. Grab the server logs when you click on that button would be helpful as well
On Fri, Mar 22, 2024 at 2:37 PM lmgithu @.***> wrote:
Thank you! Is there any way to "force" the migration again?
— Reply to this email directly, view it on GitHub https://github.com/immich-app/immich/issues/8187#issuecomment-2015784865, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGONL7XHXBUFQKRR6RHAM3TYZSB75AVCNFSM6AAAAABFDCMSJSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMJVG44DIOBWGU . You are receiving this because you commented.Message ID: @.***>
-- Alex Tran
Just to make sure, are there assets in those leftover folders or are they all empty? In any case re-running the storage migration will attempt to fix potential issues, so that is certainly worth a shot!
hi Daniel,
Unfortunately the folders are not empty, there are images in them. I run the migration job again for the original template format (yyyy-mm-dd) and it seems it reversed it to that properly, all yyyy-mm folders disappeared. So for that it seems to be working fine, but I don’t dare trying again for the yyyy-mm format again not to screw something. It would be great if somebody could check what might be the root cause so people could migrate properly.
If I can provide any further information, please let me know I will try to help.
I am somewhat certain that running the migration on yyyy-mm
again would have solved the issue as well, but of course I can't tell for sure. Many filesystem moves just happen to be flaky sometimes (most of the time actually lol).
Hey! Would be great to get some more information on what's actually happening here. If you were willing to change the storage template structure again then it would be great if you could follow these steps and then provide us with the logs for your microservices instance.
1.) Change storage template to yyyy-mm format again, let the jobs finish. 2.) Check that the assets didn't all move correctly 3.) Try running the storage template migration job again manually 4.) Provide the full logs for the microservices container after this second job completes
Also if you are able to provide an example image that fails to move, that would be great. Could you also share what filesystem you're using to store the asset files?
Cheers!
@lmgithu ?
The bug
When I set a new template for the folder structure (switching from yyyy-mm-dd to yyyy-mm only) and run the storage template migration job, I see that 1) the job finished running 2) 80% of the folders now have a mix of yyyy-mm-dd and yyyy-mm folders with data in them).
I wanted to check why immich is not capable of transforming all folders and why there are leftovers? My other contern is whether this is breaking viewing the images and/or the database.
Desired outcome:
The OS that Immich Server is running on
Debian
Version of Immich Server
v1.98.2
Version of Immich Mobile App
v1.98.2
Platform with the issue
Your docker-compose.yml content
Your .env content
Reproduction steps
Additional information
No response