[
https://issues.apache.org/jira/browse/HDDS-4904?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17297557#comment-17297557
]
Marton Elek commented on HDDS-4904:
-----------------------------------
> no production solution for CSI can be based on goofys
I think it depends on the requirements. I think goofys can be used in
production if POSIX semantics are not very important. And there are other s3
compatible fuse file systems like https://github.com/s3fs-fuse/s3fs-fuse or
https://github.com/archiecobbs/s3backer
And since HDDS-3787 we can configure the commands to mount/unmount so goofyfs
is not the only one option. That's why I would prefer to keep the documentation
of the CSI but clarify what do we provide (thin layer for CSI) and what do we
not (yet?) (native fuse drive or NFS) and where is the point where the
responsibility of Ozone ends...
> Remove mention of CSI support
> -----------------------------
>
> Key: HDDS-4904
> URL: https://issues.apache.org/jira/browse/HDDS-4904
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Arpit Agarwal
> Assignee: Dinesh Chitlangia
> Priority: Critical
>
> The Ozone website prominently mentions CSI support:
> [https://ozone.apache.org/docs/1.0.0/]
> [https://ozone.apache.org/docs/1.0.0/interface/csi.html]
> Our docs give a false impression to users that CSI is fully functional and
> supported for persistent storage inside containers.
> This support uses goofys+S3 gateway, so it is not appropriate for any serious
> usage. A real CSI solution should use an approach like the cBlocks prototype
> by building directly on top of HDDS containers with a real device driver.
> Until that time we should not claim CSI support. Alternatively we should be
> honest with our users that it is a prototype and not suitable for serious
> usage.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
