glusterfs and high availability.

[bread-baker]

I tested glusterfs over the last week and am not going to use for our high availability necessary data. It worked great until a reboot and umount caused split brain.


I'll post details if there are responses to this.

In the future I think glusterfs version 3.3+ on top of zfs will be a way to have high availability on par or better than drbd.


So we'll try to use drbd per the wiki with two resources in Primary/Primary mode.

 

[oeginc]

I tested glusterfs over the last week and am not going to use for our high availability necessary data. It worked great until a reboot and umount caused split brain.


I'll post details if there are responses to this.

In the future I think glusterfs version 3.3+ on top of zfs will be a way to have high availability on par or better than drbd.


So we'll try to use drbd per the wiki with two resources in Primary/Primary mode.

Did you use Gluster with AT LEAST 3 nodes? You need to think of gluster as a network raid-5/6 type device, it needs to be able to obtain a sane quorum in order to avoid split-brain (just like ProxMox does). It should work fine when configured for 3+ nodes, as long as it was done properly...

 

[bread-baker]

Did you use Gluster with AT LEAST 3 nodes? You need to think of gluster as a network raid-5/6 type device, it needs to be able to obtain a sane quorum in order to avoid split-brain (just like ProxMox does). It should work fine when configured for 3+ nodes, as long as it was done properly...

I followed http://docs.redhat.com/docs/en-US/Red_Hat_Storage_Software_Appliance/3.2/html/User_Guide/

and used replicated volumes per http://docs.redhat.com/docs/en-US/R...tration_Guide-Setting_Volumes-Replicated.html

The set up and user tools are nice.

But healing from problems is difficult.

Also openvz and fuse do not work at all.