SeaFile API does not delete File, although it returns "OK"

Seafile Server
1

Hi everyone,

we use the API since approx. 3 years. The overall experience is quite good.
Last week we observed some DELETE-operations (reference/delete_api-v2-1-repos-repo-id-file), that returned HTTP-status 200 (OK), but the respective files have not been deleted.

So I wanted to ask, if anybody else observed this behavior or can give a hint, how to track the origin of the problem. Unfortunately, I was not yet able to reproduce the issue.

Server Version: 11.0.17

Best
Lars

2

Just some additions:

  • We use server 11-pro on docker.
  • Some of the deleted files do not appear in the trash after deletion.
  • The problem only occurs, if we connect to the cluster address over haproxy. If we connect to one of the cluster nodes directly, the file deletion works without problems.

So maybe there is an issue with the haproxy configuration.
Any ideas?

Regards,
Dirk

3

Hello!

One question: can you confirm that haproxy is configured to pass both HTTP and WebSockets traffic? Also, did you do any updates of Seafile around when the problem started? Lastly, when this occurs, are there any corresponding errors in any logs or in the browser console?

4

We’ve made no special configuration for websockets in haproxy (never heard about that). And there were no updates performed.

5

Hello!

Ok, got it. At this point, as I mentioned above, the only idea I have is to keep trying to reproduce the issue on your end and - if & when it does - check the browser’s developer console & Seafile logs to try and glean any clues.

6

We just drained two of the three haproxy queues. Now the file deletion works without problems. We will investigate this further. The script should only open a single session to delete the files.