Content
Images lost in wiki due to worker timeouts and volume not mounted
Added by Félix Thouin 3 days ago
Hello! I am running OpenProject on a docker container self-hosted on a server. A while ago, images attached to articles in a project's wiki started disappearing!
I made a full bug report here and a wonderful person pointed out that
while trying to recreate it, also have viewed your log. It seems your worker has crashed and auto restarted with no volumes mounted. I recommend you to have a look at our documentation about the mounting or consult or forums.
I tried to look at the log (attached to this message) to find where there would be a sign of a worker crashing. I have found this at lines 5543-5546
[3574] ! Terminating timed out worker (Worker 0 failed to check in within 60 seconds): 162121
[3574] ! Terminating timed out worker (Worker 1 failed to check in within 60 seconds): 165348
[3574] - Worker 0 (PID: 168389) booted in 0.3s, phase: 0
[3574] - Worker 1 (PID: 168392) booted in 0.31s, phase: 0
however, I am not sure how to deduce from this or adjacent lines that the volumes were not mounted. Here is the docker-compose file I use to build the containers (I removed some services not related to openproject)
version: "3.7"
networks:
frontend:
backend:
services:
nginx:
ports:
- 80:80
- 443:443
image: nginx
volumes:
- ./nginx/openproject.conf:/etc/nginx/conf.d/openproject.conf
- ./nginx/wiki.conf:/etc/nginx/conf.d/wiki.conf
- ./nginx/ssl/options-ssl-nginx.conf:/etc/letsencrypt/options-ssl-nginx.conf
- ./nginx/ssl/ssl-dhparams.pem:/etc/letsencrypt/ssl-dhparams.pem
- ./nginx/ssl/fullchain.pem:/etc/letsencrypt/live/openproject.silvascience.org-0002/fullchain.pem
- ./nginx/ssl/privkey.pem:/etc/letsencrypt/live/openproject.silvascience.org-0002/privkey.pem
openproject:
stdin_open: true
tty: true
ports:
- 8080:80
environment:
- OPENPROJECT_SECRET_KEY_BASE=secret
- OPENPROJECT_HOST__NAME=openproject.silvascience.org
- OPENPROJECT_HTTPS=true
- OPENPROJECT_DEFAULT__LANGUAGE=en
image: openproject/openproject:16
volumes:
- ./openproject/pgdata:/var/openproject/pgdata
- ./openproject/assets:/var/openproject/assets
Is there anything wrong with this docker-compose file that would prevent proper remounting of the volumes?
I also notice that everytime the containers are restarted, the pgdata
and assets
owners and group permissions are changed to `drwx------ 19 tcpdump messagebus` preventing me from backing them up using the same user that launched the containers
Is that normal? Could that be related to the problem? Is there a way I could set my containers so that the user launching the containers can have access to those?
Any help is appreciated!