[pve-devel] Install glusterfs-server as default

Fabian Grünbichler f.gruenbichler at proxmox.com
Fri Aug 9 14:02:53 CEST 2019


On August 9, 2019 1:13 pm, Dominic Jäger wrote:
> As explained in detail in bug 2293 [0] glusterfsscan is not working at the moment. Our function tries to contact a nfs server that is running inside gluster servers that does not respond.
> 
> There is a command line utility simply called 'gluster' [1] that can solve this problem in a single line:
> 'gluster --remote-host=<gluster_node> volume list'
> Unfortunately, this utility is shipped in the package glusterfs-server which is not installed on PVE 6 as default.
> 
> I've now spent some hours looking for another solution but have not found anything simpler so far. Thus, I'd appreciate other ideas on this topic, especially regarding how much we want to avoid adding glusterfs-server.
> 
> [0] https://bugzilla.proxmox.com/show_bug.cgi?id=2293
> [1] https://manpages.debian.org/buster/glusterfs-server/gluster.8.en.html

it might be worth it to attempt to convince upstream to ship the 
"gluster" binary in its own package or move it to glusterfs-common 
(since it seems to be a general management CLI akin to Ceph's 'ceph', 
and not the actual server part).

I am also not sure whether one of 'peer status' or 'pool list' is not 
better suited as scan command (the man page does not give much 
information ;)). there could be lots of volumes on a gluster cluster, 
and listing them all on each scan sounds expensive.



More information about the pve-devel mailing list