Hi all, I would like to export gluster volume (which is a replica 3) with NFS so as to use for persistent container storage. I can directly use gluster storage plugin from docker containers though it seems that this approach uses FUSE mount. I have read that nfs-ganesha i using libgfapi which provides better performance thus trying to use NFS through it. I am just starting with nfs-ganesha and following the doc https://docs.gluster.org/en/latest/Administrator%20Guide/NFS-Ganesha%20GlusterFS%20Integration/ I am a bit confused with the best approach for high availability. I am already using pacemaker/corosync for other services and have a virtual IP for the cluster. I was thinking that each container can use the locally exposed NFS share (using localhost as the IP/domain of the NFS server, each container its own separate share). Reading the HA setup at the above link, it mentions the use of ganesha-ha.conf which incorporates some HA_CLUSTER_NODES etc parameters. I do not see the reason to go like that since HA is managed from gluster already. Also at same doc it mentions that this HA approach is to be replaced from storhaug which when checking at github seems like an idle repo for years. What is the best approach for HA using nf-ganesha? Thank you, Alex -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.gluster.org/pipermail/gluster-users/attachments/20201109/e8003976/attachment.html>
https://docs.gluster.org/en/latest/Administrator%20Guide/NFS-Ganesha%20GlusterFS%20Integration/ On Mon, Nov 9, 2020 at 3:08 AM Alex K <rightkicktech at gmail.com> wrote:> > https://docs.gluster.org/en/latest/Administrator%20Guide/NFS-Ganesha%20GlusterFS%20Integration/ >> Reading the HA setup at the above link, it mentions the use of > ganesha-ha.conf which incorporates some HA_CLUSTER_NODES etc parameters. > I do not see the reason to go like that since HA is managed from gluster > already. >gluster's HA is for gluster. HA for nfs-ganesha is managed by pacemaker as mentioned in the doc you referenced. The two are unrelated. storhaug is incomplete, but it works well enough for some people. YMMV. -- Kaleb -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.gluster.org/pipermail/gluster-users/attachments/20201109/800d5097/attachment.html>
Hi Alex, I have been playing arround with the NFS Ganesha on EL8 and I was surprised that the solution deploys the cluster with a 'portblock' resource which is relying on IPTABLES, when the default is NFTABLES... Also, some selinux issues came up. I think that Ganesha with Pacemaker integration is quite good and reliable and will ensure that your NFS clients will remain operational. Also, it's worth noting that RHEL support only corosync/pacemaker setup . Best Regards, Strahil Nikolov ? ??????????, 9 ??????? 2020 ?., 10:08:44 ???????+2, Alex K <rightkicktech at gmail.com> ??????: Hi all, I would like to export gluster volume (which is a replica 3) with NFS so as to use for persistent container storage. I can directly use gluster storage plugin from docker containers though it seems that this approach uses FUSE mount. I have read that nfs-ganesha i using libgfapi which provides better performance thus trying to use NFS through it. I am just starting with nfs-ganesha and following the doc https://docs.gluster.org/en/latest/Administrator%20Guide/NFS-Ganesha%20GlusterFS%20Integration/ I am a bit confused with the best approach for high availability. I am already using pacemaker/corosync for other services and have a virtual IP for the cluster. I was thinking that each container can use the locally exposed NFS share (using localhost as the IP/domain of the NFS server, each container its own separate share). Reading the HA setup at the above link, it mentions the use of ganesha-ha.conf which incorporates some? HA_CLUSTER_NODES? etc parameters. I do not see the reason to go like that since HA is managed from gluster already.? Also at same doc it mentions that this HA approach is to be replaced from storhaug which when checking at github seems like an idle repo for years.? What is the best approach for HA using nf-ganesha? Thank you, Alex ________ Community Meeting Calendar: Schedule - Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC Bridge: https://meet.google.com/cpu-eiue-hvk Gluster-users mailing list Gluster-users at gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users