5 blades are gluster bricks and KVM hosts at the same time. Gluster version: 3.8.3 OS: Arch
Everything works happily with FUSE mount and disks are defined as type file. If I change the disk definition to use libgfapi, then I can still start the VM if the disk file is local. If the disk file is located on a remote brick then I get below errors.
error: internal error: process exited while connecting to monitor: [2016-10-20 00:55:05.442560] I [MSGID: 104045] [glfs-master.c:91:notify] 0-gfapi: New graph 73736b76-6830-352d-3630-36332d323031 (0) coming up
[2016-10-20 00:55:05.442649] I [MSGID: 114020] [client.c:2356:notify] 0-vstore-client-1: parent translators are ready, attempting connect on transport
[2016-10-20 00:55:05.445718] I [MSGID: 114020] [client.c:2356:notify] 0-vstore-client-2: parent translators are ready, attempting connect on transport
[2016-10-20 00:55:05.446194] I [rpc-clnt.c:1947:rpc_clnt_reconfig] 0-vstore-client-1: changing port to 49153 (from 0)
[2016-10-20 00:55:05.448419] I [MSGID: 114020] [client.c:2356:notify] 0-vstore-client-3: parent translators are ready, attempting connect on transport
[2016-10-20 00:55:05.449343] I [MSGID: 114057] [client-handshake.c:1446:select_server_supported_programs] 0-vstore-client-1: Using Program GlusterFS 3.3, Num (1298437), Version (330)
[2016-10-20 00:55:05.449731] I [rpc-clnt.c:1947:rpc_clnt_reconfig] 0-vstore-clien
Cluster is set up as distribution. So if physical disk file location and the VM are on the same machine - no problems. I get the error otherwise.
Any idea why it could be happening?
qemu user and group id = 0 gluster volume user and group id = 0
option rpc-auth-allow-insecure on
Thanks
