[Bugs] [Bug 1355956] New: RFE : move ganesha related configuration into shared storage

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1355956

            Bug ID: 1355956
           Summary: RFE : move ganesha related configuration into shared
                    storage
           Product: GlusterFS
           Version: mainline
         Component: ganesha-nfs
          Assignee: jthottan at redhat.com
          Reporter: jthottan at redhat.com
                CC: bugs at gluster.org



Description of problem:
In the present design, we maintain a copy of the below database (required for
NFS-Ganesha setup) on all the nodes of the Gluster Storage pool -
* /etc/ganesha/ganesha-ha.conf (with the details about the nodes forming
NFS-Ganesha cluster and their VIPs to be configured)

* /etc/ganesha/ganesha.conf (contains configuration parameters and the volumes
list to be exported, used by NFS-Ganesha process while starting)

* /etc/ganesha/exports/ganesha.$vol.conf (config file of a volume created when
a volume is being exported by NFS-Ganesha)

* /etc/ganesha/exports/.export_added (each volume should have unique ExportID. 
This file is to keep track of the last ExportID used)

Outstanding Issues:
Since each node shall have a local copy, we have few issues with respect to
synchronizing them across the nodes. Few of them are outlined below: 
* When a node reboots, depending on the order the services ‘pacemaker’ and
‘glusterd’ starts, we may run into a  case, where in glusterd couldn’t sync the
export files on this node.

* There could be case where in ‘.export_added’ could go out of sync across the
nodes thus resulting in different ExportIDs for the same volume exported across
different nfs-ganesha heads. This shall result in NFS mounts throwing “Stale
File handle error” post failover 

* When a node is down while performing “refresh-config”, the export config of
that volume doesn’t get synced when it comes back up.

* As one of the pre-requisites, ganesha-ha.conf has to be copied to all the
nodes in the gluster cluster before enabling nfs-ganesha (setup).

* When any changes are made to main “/etc/ganesha.conf” file, it has to be
manually copied to all the nodes in the cluster.

* When “refresh-config” is performed, the script syncs up the volume config 
only across the NFS-Ganesha cluster but not to all the nodes in the Gluster
storage pool

Version-Release number of selected component (if applicable):
mainline


Additional info:
The existing users shall be impacted while doing upgrade. We may need to
provide scripts for the smooth upgrade to move the older config data to
shared_storage and also few of the setup pre-requisites need to be
removed/modified accordingly.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=3i9R2hbZQy&a=cc_unsubscribe