linux-ha-storage / storhaug Goto Github PK

https://bugzilla.redhat.com/show_bug.cgi?id=1461098

Issue was fixed by adding new SELinux boolean. (using this boolean you can allow some actions, in this case allow ganesha to acces fusefs_t labels.)

If you would like to switch this boolean on during installation ganesha you can follow this tutorial:
https://mojo.redhat.com/docs/DOC-1131234-creating-own-product-policies#jive_content_id_61_Setting_Booleans_During_a_Product_Policy_Installation

We should allow for cluster admins to use their own STONITH devices for fencing.

http://review.gluster.org/14878 was merged relatively recently and will need to be added to Storhaug too. This prevents clients from getting stuck in the TCP-timeout which can last up to 20 minutes.

Look into using Pacemaker Remote for either the access methods (e.g. SMB, NFS), the storage daemons (e.g. GlusterFS), or both.

There are a few functions in the main storhaug script that are marked TODO for moving into the Ganesha RA. Move them.

Hello,

GlusterFS 3.12 and up is using storhaug instead of gluster nfs.ganesha commands, but sadly, there isn't any documentation how to actually use it. Would you mind adding a simple document how to use it please? I'll give an example scenario:

• 2 nodes, on each one there are 2 drives, 1st for the OS and the 2nd for glusterfs. 2 NICS on each node - 1 for internal communication for nodes, other NIC for serving content from nfs-ganesha.
• Gluster will be created for a simple replication (HA) for 2 nodes
• NFS Ganesha will be installed, and will be used along with pcs and a cluterIP.

Where in this scenario does storhaug goes in and what should I type? that sort of document would help us a lot (you can see the frustration of a missing documentation on the gluster-user mailing list).

Another question: looking at the scripts, you're trying to parse ganesha.conf. Why? it should only include an "include" line for the "exports" directory in /etc/ganesha/exports/

Final question: with storhaug, do I need to create ganesha-ha.conf file or does storhaug creates it?

Thanks

Create a Pacemaker resource agent (RA) for storhaug. This RA would take care of managing the storhaug configuration across all cluster nodes (e.g. config reload) and reduce the number of parameters in the config files. This should also allow for storhaug to integrate into existing Pacemaker clusters.

As a followup to #1, improve the portblock RA to use conntrack to track TCP connections for tickling rather than flat files populated via netstat.