0

We have cluster with CephFS configured. It is Bluestore-based and uses SSD-based OSDs for metadata pool for performance reasons. We want to add access to files on this FS via S3. Is it possible to use Ceph Object Gateway for this, or it will not be able to control the same OSDs/pools which are under CephFS control? I see that sometimes people tries to solve opposite problem (mount Object Storage as FS, using s3fs), but for us CephFS is the primary thing and S3 access to is just add-on (for historical reasons we use minio for this now, but looking for Ceph-native alternative). Here I see that author wanted to use CephFS to access files created via Object Gateway, this is almost those what we want (with exception that priorities are different (for us CephFS is on the first place)). Answer there is not promising, but I wonder if something has been changed since then?

1 Answers1

0

The answer came from Daniel Persson on youtube comments section (https://youtu.be/6uX23Q3y_SU):

No, CephFS and S3 use the same backend, Rados Object Storage. In this storage solution, you store multiple objects spread over different placement groups in a cluster of object storage devices. But the pools you set up for each solution are separate, and the two implementations are strictly different even though they use the same backend.