[samba&ctdb]A question about a samba cluster witch ctdb

[guoleia at 139.com]

Hi folks,

I'm using ctdb for a samba cluster.
ctdb version:1.0.114-1
smbd version 3.5.6
Linux server:suse10

There are 4 nodes(eg. nodesA/nodeB/nodeC/nodeD) in the cluster.

I'm using a win2003 server as samba client to write and read lots files on nodeA.During the period,since the nodeB is powered down,I find that the data transfering between win2003 and nodeA is hanging on nearly 20~30 seconds.Besides,I capture  no any data packets  between win2003 and nodeA  with WIRESHARK in that 20~30 seconds.


View log.ctdb, find the following:
**********************************************************************************************************************************
  39076 2011/12/28 11:59:38.109833 [ 5605]: client/ctdb_client.c:771 control timed out. reqid:36158 opcode:91 dstnode:2
  39077 2011/12/28 11:59:38.109888 [ 5605]: client/ctdb_client.c:882 ctdb_control_recv failed
  39078 2011/12/28 11:59:38.109902 [ 5605]: client/ctdb_client.c:1225 ctdb_control for getnodes failed ret:-1 res:-1
  39079 2011/12/28 11:59:38.109912 [ 5605]: server/ctdb_recoverd.c:952 Unable to get nodemap from remote node 2
  39080 2011/12/28 11:59:38.109928 [ 5605]: Unable to update local flags
  39081 2011/12/28 11:59:47.122383 [ 5605]: Dropped orphaned reply control with reqid:36158
  39082 2011/12/28 11:59:47.122418 [ 5605]: ctdb_control error: 'node is disconnected'
  39083 2011/12/28 11:59:47.122432 [ 5605]: client/ctdb_client.c:1225 ctdb_control for getnodes failed ret:-1 res:-1
  39084 2011/12/28 11:59:47.122443 [ 5605]: server/ctdb_recoverd.c:952 Unable to get nodemap from remote node 2
  39085 2011/12/28 11:59:47.122455 [ 5605]: Unable to update local flags
**********************************************************************************************************************************


I guess it due to the state and data synchronization among the the rest 3 nodes.
I would like to know ,is it correct in this case?Is there any ways to resolv?