[Gluster-users] gluster peer probe

[Piotr Skurczak]

Hello everyone,

So this is my first email here. Recently I have downloaded glusterfs-3.1 and
tried to install it on my four servers. I did not have any problems with the
installation, but the configuration is a little bit unclear to me. I would
like to ask you, if perhaps you had the same problems as I encountered and
if yes, then how did you resolve your problem ?

My configuration : 5 servers scattered around the world globe : 2 x VPS in
the UK and USA, 1 (fully) virtual server (with bridged eth adapter on
account of that my internal ifconfig shows another ip, and the public ip is
different, but the server is pingable, and ports are accessible too), and 2
servers behind routers in so called "demilitarized zone".

Since glusterfs does not need any file system (at least it was never
mentioned in installation instructions) I decided to set up a catalogue
(/data) where I am supposed to keep my data and replicate the data onto the
other /data folders on the other servers.

So we have 5 ip addresses that look like :

83.31.111.111
93.20.112.120
(...)

All ip's can ping one another. I can do ssh from one onto another. I checked
iptables rules to allow gluster on communication on whatever ports it need
Now what I do is "gluster ping probe HOSTNAME", and this is where it
either stuck, or says that it failed, or gives strange error logs.

This is an excerpt from gluster logfile :

[2011-01-09 14:59:34.273183] I
[glusterd-handler.c:674:glusterd_handle_cli_list_friends] glusterd: Received
cli list req
[2011-01-09 14:59:44.896781] I
[glusterd-handler.c:716:glusterd_handle_cli_get_volume] glusterd: Received
get vol req
[2011-01-09 15:16:16.11424] I
[glusterd-handler.c:2387:glusterd_handle_probe_query] glusterd: Received
probe from uuid: 637b41c6-3349-4202-8cab-2d1293de782c
[2011-01-09 15:16:16.11472] I [glusterd-handler.c:386:glusterd_friend_find]
glusterd: Unable to find peer by uuid
[2011-01-09 15:16:16.443504] I [glusterd-handler.c:398:glusterd_friend_find]
glusterd: Unable to find hostname: 93.20.112.120
[2011-01-09 15:16:16.443526] I
[glusterd-handler.c:2401:glusterd_handle_probe_query] glusterd: Unable to
find peerinfo for host: 93.20.112.120 (24007)
[2011-01-09 15:16:16.444769] W [rpc-transport.c:849:rpc_transport_load]
rpc-transport: missing 'option transport-type'. defaulting to "socket"
[2011-01-09 15:16:16.447238] I [glusterd-handler.c:2600:glusterd_friend_add]
glusterd: connect returned 0
[2011-01-09 15:16:16.447278] I
[glusterd-handler.c:2422:glusterd_handle_probe_query] glusterd: Responded to
83.31.111.111, op_ret: 0, op_errno: 0, ret: 0
[2011-01-09 15:16:16.491402] E [socket.c:1656:socket_connect_finish]
management: connection to  failed (No route to host)
[2011-01-09 15:16:16.491726] I
[glusterd-handler.c:2131:glusterd_handle_incoming_friend_req] glusterd:
Received probe from uuid: 637b41c6-3349-4202-8cab-2d1293de782c
[2011-01-09 15:16:16.491751] I [glusterd-handler.c:386:glusterd_friend_find]
glusterd: Unable to find peer by uuid
[2011-01-09 15:16:16.491767] I
[glusterd-utils.c:2101:glusterd_friend_find_by_hostname] glusterd: Friend
93.20.112.120 found.. state: 0
[2011-01-09 15:18:40.413343] I
[glusterd-handler.c:674:glusterd_handle_cli_list_friends] glusterd: Received
cli list req
[2011-01-09 15:18:40.413449] I
[glusterd-handler.c:2768:glusterd_xfer_friend_add_resp] glusterd: Responded
to 83.31.111.111 (0), ret: 0
[2011-01-09 15:18:40.413639] I
[glusterd-utils.c:2101:glusterd_friend_find_by_hostname] glusterd: Friend
93.20.112.120 found.. state: 2

On all of my servers I can do telnet HOST 24007, and it works. glusterd
process is up and running on all servers.

What I did was that I added my own names into /etc/hosts like node1 node2
node3... did not help.
There is an interesting video over here :
http://www.youtube.com/user/GlusterStorage#p/u/3/sSCCZLzNnUQ where gluster
guys just hit enter and bah, a peer is in a cluster. Reality, however, tells
another story. It is either the instruction that is incomplete or does not
consist of all tips and tricks OR I am just missing something.

Any help in this matter would be highly appreciated.
Peter