Hi. I'm using glusterfs 3.4 in replica 2 configuration. My main usage of gluster is for openstack storage backend (glance and shared storage for nova-compute) and i'm using gluster and i'm using gluster for about 1 year (not really a production environment, so a limited set of experiments is possible). I have some doubts that i want to share with the users mailing list:
1) Glusterfs client or nfs client? I think is better the glusterfs client. But i would like that the more experienced users give a precise answer....
2) How glusterfs manage backup-volfile-servers? My answer: from the point of view of a client (glusterfs client), if we are using this parameter the client is not affected from a down of a storage server in the replica pool. But the problem is that if a client writing a file when the problem happens, i think this file will be damaged. How a glusterfs client manage this case? It will drop the tcp connection to "storage1" after a (configurable) time, and opens a new tcp connection to "storage2". But the clients has some sort of "buffer" for data that uses during the time necessary for close and open a new tcp connection? Help please....
3) If one server goes down in a replica pool, how can i recover and re-align the data on this server? I think i should use the healing procedure. But also in this case, experienced users are welcome.... This is a particular case, but the more general case is: how can i reboot a server in a replica configuration?
4) It is better filesystem storage or block storage for openstack? This question seems to be off topic, but glusterfs has an integration for cinder. So block storage or file storage is an important decision and also a possibility (imho)
5) How can i do a monitor on the disks state? Glusterfs has a command for this? Should i use "external" tools?
6) How can i manage the case when a brick is damaged/corrupted? I think there is a different behavior by glusterfs version. It seems that 3.5 can manage this case. Please, an help here!
7) How can i add a new partition with xfs? Should i reboot the server? If i reboot the server, i (re) jump to the point 3. This is an important question because i don't know if tomorrow i need a new partition!
These are my doubts. But these are doubts very commons between who plans to use glusterfs. So, i think that if this mail will have useful responses, it could be a reference for new glusterfs users.
Thank you.
MM
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://www.gluster.org/mailman/listinfo/gluster-users
