RP/MSQ 4.0 (UNIX, NT) - xgroup reconnect link issues with failover entries (Doc ID 770883.1)

Last updated on JANUARY 16, 2023

Applies to:

Goal

MessageQ for UNIX, V5.0-RP22   (main node)
MessageQ for UNIX, V4.0A-RP48 (failover nodes)

Operating systems:  Main server is Solaris 8. Failover servers are Solaris 2.5.1 and HP-UX 10.20

Problem with failover entries in the xgroup table.  To reproduce this, I have three machines with the following
configurations:

1) <NODE5>- running MessageQ, V5.0-RP22, group GRP1
2) <NODE3>- running MessageQ, V4.0A-RP48, group GRP2
3) <NODE1>- running MessageQ, V4.0A-RP48, group GRP3

Only one copy of group 2007 runs at a time.

The group.ini and set up files are under:

     ../cases/mq/<> - for <NODE5>
     ../cases/mq/<> - for <NODE2>
     ../cases/mq/<> - for <NODE3>

Thinking that RP50 may have an impact on R47, I tried this with RP50 as well on the failover remote nodes, and was
still able to reproduce this using<NODE5> (V5.0A-RP22) as well as V4.0A-RP50 on <NODE2>
(../cases/mq/<>).  I had to use<NODE2> , since no patches beyond RP48 were available for Solaris
and HP-UX.

1)Start<NODE2>
start <NODE5>
Will connect to<NODE2>
Bring down<NODE2>, bring up within 2 minutes
Will connect to<NODE2>
Bring down<NODE2>; bring up after several minutes (> 2 minutes)
Will not connect to<NODE2>

2) Shutdown both remote groups
Start <NODE5>
After several minutes (> 2 minutes) start<NODE2>
Will not connect to <NODE5>

Note:  these links can be started manually through dmqmonc, but is not acceptable to the customer who depends on
automatic reconnect. If there is no failover entry for 2007 in the group <GROUP1> xgroup table, link seems to recover
automatically.

Solution

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.