Bricks on NFS

[dlucredativ]

There is an old thread [1] about placing bricks on NFS. It was not possible at that time, due to unsupported extended attributes over NFS. Today, NFS4 supports extended attributes in the user namespace [2]. Unfortunately, gluster uses the trusted namespace [3].

Are there any plans to make gluster work in the user namespace?

[1] https://gluster-users.gluster.narkive.com/XGWL0P6q/can-i-create-a-gluster-volume-from-nfs-mounted-partitions
[2] https://datatracker.ietf.org/doc/html/rfc8276#page-7
[3]

glusterfs/libglusterfs/src/glusterfs/glusterfs.h

Lines 95 to 109 in c22bf0f

	#define GF_XATTR_CLRLK_CMD "glusterfs.clrlk"
	#define GF_XATTR_INTRLK_CMD "glusterfs.intrlk"
	#define GF_XATTR_PATHINFO_KEY "trusted.glusterfs.pathinfo"
	#define GF_XATTR_NODE_UUID_KEY "trusted.glusterfs.node-uuid"
	#define GF_XATTR_LIST_NODE_UUIDS_KEY "trusted.glusterfs.list-node-uuids"
	#define GF_REBAL_FIND_LOCAL_SUBVOL "glusterfs.find-local-subvol"
	#define GF_REBAL_OLD_FIND_LOCAL_SUBVOL "glusterfs.old-find-local-subvol"
	#define GF_XATTR_VOL_ID_KEY "trusted.glusterfs.volume-id"
	#define GF_XATTR_LOCKINFO_KEY "trusted.glusterfs.lockinfo"
	#define GF_META_LOCK_KEY "glusterfs.lock-migration-meta-lock"
	#define GF_META_UNLOCK_KEY "glusterfs.lock-migration-meta-unlock"
	#define GF_XATTR_GET_REAL_FILENAME_KEY "glusterfs.get_real_filename:"
	#define GF_XATTR_USER_PATHINFO_KEY "glusterfs.pathinfo"
	#define GF_INTERNAL_IGNORE_DEEM_STATFS "ignore-deem-statfs"
	#define GF_XATTR_IOSTATS_DUMP_KEY "trusted.io-stats-dump"

[aravindavk]

Link to previous discussion: #1239 (comment)

@amarts ^^