Open mrrodge2020 opened 2 years ago
I'm running the same version (1.5.0) also in Docker. I'm running vsftpd, but this may also apply to SMB. Make sure your consume directory has proper permissions set on the consume dir and files. I was seeing the exact same logs and noticed that the ftp service was writing files to the consume directory with only "rw" permissions as the "ftp" account I set up.
I was able to fix this by running the "setfacl" command suggested here: https://askubuntu.com/questions/969056/make-files-uploaded-by-vsftpd-automatically-inherit-owner-from-parent-directory
I ran: sudo setfacl -R -d -m u:my_root_docker_user:rwx /path/to/consume/dir
Replace "my_root_docker_user" with the username of your root account.
If that works for you, it may be a worthy addition to the FTP documentation for paperless.
Side note: I have no idea what the security implications of adding "x" are in a public/shared environment. I am only running paperless and the ftp service on a private lan, so I'm ok with it for my situation.
Thanks - Most of it straight over my head though. I'm running a Windows SMB share with full control permissions. Docker host is running on Debian and is mapped to a folder with 1000:1000 ownership. Nothing seems to have any access trouble, it's just the queue not starting up after an action gets postponed.
Ah. My docker host and ftp server are on the same ubuntu box. That's why a lot of it probably didn't make sense.
Ah right haha. I genuinely don't think this is a permission issue as other docs process OK - it's because the scanner is still writing the file when paperless picks it up, so it gets added to the queue to check back later, which it then never does.
I feel like I'm having a similar problem, but I'm not seeing Paperless even identifying the files are there (i.e. no Waiting for file
log entry.) I have the polling set to 30 so I don't know why its not getting picked up. I have Paperless running in Docker on a Raspberry Pi 4 with a SMB folder mounted for the consume folder.
Ah right haha. I genuinely don't think this is a permission issue as other docs process OK - it's because the scanner is still writing the file when paperless picks it up, so it gets added to the queue to check back later, which it then never does.
This is quite a useful hint, since I believe that this problem might explain an observed problem which prevents Paperless to process large PDF files (~30 pages)
Describe the bug In-use files (still being written by scanner but discovered by paperless) state they're added to the task queue but then never get processed. Other documents get discovered and processed after this and the stalled documents get processed after a reboot.
To Reproduce Scan a document to an SMB share for consumption. Ensure it gets discovered by paperless before the scanner has finished writing the file.
Expected behavior Would expect the task queue to process the file later, allowing time for the scanner to release the file.
Screenshots N/A
Webserver logs
Relevant information
-docker-compose.yml: