[Pgcluster-general] PgCluster1.9.0rc3

[Lia Domide]

Hi all,

 

I am testing pgCluster's new release 1.9.0rc3, in a two nodes environment,
with Ubuntu7.04.

It seems to work fine, almost.

The following scenario fails:

Start DB1, DB2 (DB1 runs on pg2 host, and DB2 runs on pg3 host in the below
log)

Start REP

Start LB

            -- OK: do some inserts now.

Stop DB2

P1: sometimes OK: do some inserts now.

P2: Some other time BAD: Trying to connect to LB hangs, having the same log
message as below.

Start DB2 -R

Stop DB1 

Stop LB

Start LB

            -- BAD: Trying to connect to LB, hangs . See the below log for
LB.

Stop REP

Start REP

            -- OK now it works, if P1 before.

 

2008-03-11 13:11:33 [17376] DEBUG:PGRset_status_on_cluster_tbl():host:pg2
port:5432 max:32 use:0 status1
2008-03-11 13:11:33 [17376] DEBUG:PGRset_status_on_cluster_tbl():host:pg3
port:5432 max:32 use:0 status1
2008-03-11 13:11:33 [17376] DEBUG:init_pglb():Child_Tbl size is[49536]
2008-03-11 13:11:35 [17376] DEBUG:PGRscan_cluster:2 ClusterDB can be used
2008-03-11 13:11:35 [17376] DEBUG:PGRscan_cluster:pg2 [5432],useFlag->1
max->32 use_num->0
2008-03-11 13:11:35 [17376] DEBUG:PGRset_status_on_cluster_tbl():host:pg2
port:5432 max:32 use:1 status2
2008-03-11 13:11:35 [17381] DEBUG:I am 17381
2008-03-11 13:11:35 [17381] DEBUG:I am 17381 accept fd 6
2008-03-11 13:11:35 [17381] DEBUG:Protocol Major: 3 Minor: 0 database: TEST
user: postgres
2008-03-11 13:11:35 [17381] ERROR:connect_inet_domain_socket: connect()
failed: Connection refused
2008-03-11 13:11:35 [17381] DEBUG:PGRset_status_on_cluster_tbl():host:pg2
port:5432 max:32 use:2 status98
2008-03-11 13:11:35 [17376] DEBUG:PGRset_status_on_cluster_tbl():host:pg2
port:5432 max:32 use:98 status99
User defined signal 2
$ 2008-03-11 13:11:35 [17381] DEBUG:child receives close connection request

 

If REP service is also restarted, LB is no longer hanging this way. 

I cannot describe yet why sometimes the LB does not succeed connecting to
the live DB node, when the other is down. Does anyone have any idea?  

 

                                        Cheers,

                                            Lia.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgcluster-general/attachments/20080311/c56f3e05/attachment.html