[Gluster-devel] AFR Replication

[Christopher Hawkins]

I confirmed the same behavior in 1.3.7 built from source and upgraded to
latest tarball as per Krishna. At first I had the same experience as Gordan
but might have found the bug - 

When using a config that specifies ip auth as follows on node A:
option auth.ip.cluster-export.allow 192.168.*   # Allow access to
"cluster-export" volume
option auth.ip.cluster-export.allow 127.0.0.1

I have all kinds of problems on node B. ls takes forever, writes do not
sync, and I see the following the glusterfs log: 

2008-04-18 08:15:23 E
[client-protocol.c:5089:client_protocol_handshake_reply] cluster-remote:
SETVOLUME on remote-host failed: ret=-1 error=Authentication failed
2008-04-18 08:15:23 W [client-protocol.c:279:client_protocol_xfer]
cluster-remote: attempting to pipeline request type(1) op(19) with handshake
2008-04-18 08:15:23 W [client-protocol.c:4759:client_protocol_cleanup]
cluster-remote: cleaning up state in transport object 0x86bf9e8
2008-04-18 08:15:23 E [client-protocol.c:4809:client_protocol_cleanup]
cluster-remote: forced unwinding frame type(1) op(19) reply=@0x86f8780
2008-04-18 08:15:23 E [client-protocol.c:4025:client_setxattr_cbk]
cluster-remote: no proper reply from server, returning ENOTCONN
2008-04-18 08:15:23 C [afr.c:5795:afr_check_xattr_cbk] afr: [CRITICAL]:
'cluster-remote' doesn't support Extended attribute
2008-04-18 08:15:24 E
[client-protocol.c:5089:client_protocol_handshake_reply] cluster-remote:
SETVOLUME on remote-host failed: ret=-1 error=Authentication failed
2008-04-18 08:15:24 W [client-protocol.c:4759:client_protocol_cleanup]
cluster-remote: cleaning up state in transport object 0x86bf9e8

However, when I specify auth like this (as I am on node B):
option auth.ip.cluster-export.allow *

Then node A is quite happy. The logs show no errors and writes are
immediately mirrored. When I set both servers to use this kind of auth, the
mirror works in both directions. Are we looking at an auth bug?

Thanks, 
Chris 


-----Original Message-----
From: gluster-devel-bounces+chawkins=veracitynetworks.com at nongnu.org
[mailto:gluster-devel-bounces+chawkins=veracitynetworks.com at nongnu.org] On
Behalf Of gordan at bobich.net
Sent: Friday, April 18, 2008 7:31 AM
To: Gluster Devel
Subject: Re: [Gluster-devel] AFR Replication

On Fri, 18 Apr 2008, Krishna Srinivas wrote:

> If you still get errors, check for error messages when you start the 
> client.

OK, upgraded to 1.3.8pre5 RPMs. Started both servers, then both clients. 
Now the logs say:

glusterfs.log:
2008-04-18 12:22:18 C [afr.c:6185:afr_check_xattr_cbk] foo: 
[CRITICAL]: 'foo1' doesn't support Extended attribute
2008-04-18 12:22:18 E
[client-protocol.c:5089:client_protocol_handshake_reply] foo2: 
SETVOLUME on remote-host failed: ret=-1 error=Authentication failed
2008-04-18 12:22:18 W [client-protocol.c:279:client_protocol_xfer]
foo2: attempting to pipeline request type(1) op(19) with handshake
2008-04-18 12:22:18 W [client-protocol.c:4759:client_protocol_cleanup]
foo2: cleaning up state in transport object 0x60ee60
2008-04-18 12:22:18 E [client-protocol.c:4809:client_protocol_cleanup]
foo2: forced unwinding frame type(1) op(19) reply=@0x612330
2008-04-18 12:22:18 E [client-protocol.c:4025:client_setxattr_cbk]
foo2: no proper reply from server, returning ENOTCONN
2008-04-18 12:22:18 C [afr.c:6185:afr_check_xattr_cbk] appleseed: 
[CRITICAL]: 'foo2' doesn't support Extended attribute
2008-04-18 12:22:18 C [tcp.c:87:tcp_disconnect] foo2: connection
disconnected

glusterfsd.log:
2008-04-18 12:22:14 E [authenticate.c:173:gf_authenticate] auth: Nobody
cares to authenticate!! Rejecting the client foo
2008-04-18 12:22:14 E [server-protocol.c:5633:mop_setvolume] server: 
Cannot authenticate client from 192.168.0.1:1021
2008-04-18 12:22:14 E [server-protocol.c:6002:server_protocol_interpret]
server: bound_xl is null
2008-04-18 12:22:14 C [tcp.c:87:tcp_disconnect] server: connection
disconnected
2008-04-18 12:22:15 E [protocol.c:271:gf_block_unserialize_transport]
server: EOF from peer (192.168.0.1:1020)
2008-04-18 12:22:15 C [tcp.c:87:tcp_disconnect] server: connection
disconnected
2008-04-18 12:22:18 W [posix.c:1511:posix_setxattr] foo: setxattr on
/: Operation not supported

Does this shed more light on what is going wrong?

Gordan


_______________________________________________
Gluster-devel mailing list
Gluster-devel at nongnu.org
http://lists.nongnu.org/mailman/listinfo/gluster-devel