Hi Aurélien,
I can provide you an alternate solution, still nfs-ganesha based. For
the need of a project, I developed an open-source library that emulate a
POSIX namespace using a KVS (for metadata) and an object store (for
data). For example, you can use REDIS and RADOS. I have written a FSAL
Hello Philippe,
Did you mean it like I could directly use the FSAL you developed, modify
some code (or not?) in order to use S3 as a storage? Or is it to share
solutions you found to some problems I will encounter during the
development of my FSAL_S3.
In both cases, I sure would be happy to have
That would work.
The logic that the RGW FSAL uses to do all of Frank describes is in
the Ceph source rather than FSAL_RGW for the same reason. The
strategy I take is to represent the RGW file handle as concatenated
hashes of S3/swift container name and full object name, which yields
stable handle
>From Girjesh Rajoria :
Girjesh Rajoria has uploaded this change for review. (
https://review.gerrithub.io/393642
Change subject: FSAL_GLUSTER: Add support for SEEK procedure
..
FSAL_GLUSTER: Add support for SEEK procedure
Th
I did some work with my script for extracting patch titles and change ids
and with some manual work on the output, produced the following list of
patches not included in V2.5-stable (the V2.6 tags are included to help
identify when the patches arrived):
> I0ccf28339b9296115520a5d54538f02df5ee0089
Hmm, you listed patches from very old tags as well. V2.5 should have only
patches that fix defects. We should NOT be adding new features, clean up
works etc from V2.6 to V2.5-stable. For example, you listed gerrirt
change-id . It implemented a new
feature and it is buggy as well, so no point in tak
