feat(deps): bump seaweedfs to 4.29 and migrate into weed mini#4345
feat(deps): bump seaweedfs to 4.29 and migrate into weed mini#4345aldy505 wants to merge 5 commits into
weed mini#4345Conversation
Should help with executing lifecycle commands here: #4263
Can you please mention what feature from 4.17 to 4.25 helps here? |
![sentry[bot]](https://avatars.githubusercontent.com/in/12637?s=60&v=4){kind=small}
| -volume.port=8080 | ||
| -volume.port.grpc=18080 | ||
| -ip.bind=0.0.0.0 | ||
| -webdav=false |
There was a problem hiding this comment.
Bug: The seaweedfs health check uses port 9340, but the service command doesn't configure this port. The service may not be listening on 9340, causing the health check to fail.
Severity: HIGH
Suggested Fix
Verify the correct health check port for the SeaweedFS 4.29 weed mini volume server. If 9340 is incorrect, update the health check URL to use the correct port, such as the previously used 8080 or another documented default. If 9340 is correct, add an explicit port mapping and/or command-line flag to the seaweedfs service definition to ensure it listens on that port.
Prompt for AI Agent
Review the code at the location below. A potential bug has been identified by an AI
agent. Verify if this is a real issue. If it is, propose a fix; if not, explain why it's
not valid.
Location: docker-compose.yml#L282
Potential issue: The health check for the `seaweedfs` service is configured to make a
request to `http://seaweedfs:9340/healthz`. However, the `weed mini` command for the
service does not include any flags to expose or bind to port `9340`. The previous
configuration explicitly used port `8080`. While some external documentation hints that
the Volume Server might use port `9340`, the official SeaweedFS documentation for `weed
mini` does not list it as a default port. This mismatch will likely cause the health
check to fail, marking the container as unhealthy and potentially causing dependent
services to fail to start.
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 15595c7. Configure here.
| -filer.defaultReplicaPlacement=000 | ||
| -master=true | ||
| -master.port=9333 | ||
| -master.port.grpc=19333 |
There was a problem hiding this comment.
Removed -volume.max=0 may limit volume count
Medium Severity
The old configuration explicitly set -volume.max=0 which auto-configures volume count based on available disk space. This flag was removed in the switch to mini mode. In the server command, the default for volume.max is 8 (not unlimited). If the mini command inherits this same default, the system will be limited to 8 volumes, after which new writes will fail — potentially causing data loss for nodestore and profiles storage.
Reviewed by Cursor Bugbot for commit 15595c7. Configure here.
|
Okay that works? |
aldy505
changed the title
aminvakil
left a comment
aminvakil
left a comment
There was a problem hiding this comment.
I agree with this, but maybe we should change PR's name that is not just about bumping version, but weed server is getting changed to weed mini.
aldy505
changed the title
weed mini
aldy505
changed the title
weed miniweed mini
|
What is the status? |
@teankie honestly, I'm not brave enough to merge this. Still need some tests from self-hosted users. |
We've been running this setup without issues for a few weeks now. I did change some things but nothing major. When I'm back at work tomorrow I can share the changes to the compose we made and whether they were critical changes or not. |
@dannypas00 We (at Siemens) would also be interested in your feedback. We are testing this fix as well on our stage system since 1-2 days now. The main difference is that I went with the We do have some warnings (grpc timeouts sometimes) but our 90-day retention policy was successfully applied to nodestore/profiles. However, since we accumulated a lot of data (400+ GB), it's hard to see the vacuum/cleanup effect yet. Our plan is to keep monitoring the volume status/how the storage evolves over the next days. /cc @dlouzan @max-wittig |
|
Whoa I didn't expect people would try this patch. Thank you! I personally prefer to be a few versions behind, so that I can understand what needs to be changed after this version, and catch bug early. |
|
I can now confirm, we at Fotocadeau have been running this exact diff for three weeks without any issues so far. No errors in the logs either. We didn't remove the seaweedfs worker or admin containers, we only changed the seaweedfs container itself in the .override.yml compose file. |
6 participants
Should help with executing lifecycle commands here: #4263