|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
Cannot start up the cluster channel by using hostname |
« View previous topic :: View next topic » |
Author |
Message
|
gbaddeley |
Posted: Wed Sep 30, 2009 5:00 pm Post subject: |
|
|
 Jedi Knight
Joined: 25 Mar 2003 Posts: 2538 Location: Melbourne, Australia
|
bruce2359 wrote: |
From the WMQ Intercommunications manual:
If the transport type is TCP
This is either the hostname or the network address of the remote machine
(or the local machine for cluster-receiver channels). For example,
(MACH1.ABC.COM), (fe80:43e4:0204:acff:fe97:2c34:fde0:3485) or
(19.22.11.162). It may include the port number, for example
(MACHINE(123)). It can include the IP_name of a z/OS dynamic DNS group
or a network dispatcher input port. |
Actually, with clusrcvr channels, a host name in a conname needs should be resolvable by all other hosts in the cluster, as the qmgrs on those hosts may use it to automatically define a clussdr channel and then start it. _________________ Glenn |
|
Back to top |
|
 |
karmen |
Posted: Wed Sep 30, 2009 7:32 pm Post subject: |
|
|
Newbie
Joined: 29 Sep 2009 Posts: 9
|
I know I should not restart the QM in order to make it worked.
But I was so confused that when I defined the channel with the hostname as the conname, the channel cannot start. But when I stopped the channel, altered the hostname to IP and then started the channel again, it can connect successfully.
So what's wrong?
Is the MQ cannot found the hostname from the host table? |
|
Back to top |
|
 |
fjb_saper |
Posted: Wed Sep 30, 2009 7:34 pm Post subject: |
|
|
 Grand High Poobah
Joined: 18 Nov 2003 Posts: 20756 Location: LI,NY
|
karmen wrote: |
Hi all,
I have tested; the configuration that I have posted was workable.
When I restarted the QM - MQM1 in z/OS again, the status of both cluster-sender channels (‘TO.MQMAQ01’ & ‘TO.MQMAQ07’) became running even hostname is being used in the CONNAME field.
If all the channel definitions are correct, they should be able to work automatically. But why it doesn't? Why I need to restart the QM in order to make it work? |
I don't understand much about zOS...
TCP/IP setup is I understand different under zOS from distributed machines (AIX). Could it be that some set up was made that was not picked up until the qmgr restarted reinitializing stuff like chinit space etc...?
You'll have to wait until somebody with solid zOS knowledge can comment.
Have fun  _________________ MQ & Broker admin |
|
Back to top |
|
 |
bruce2359 |
Posted: Thu Oct 01, 2009 8:02 am Post subject: |
|
|
 Poobah
Joined: 05 Jan 2008 Posts: 9470 Location: US: west coast, almost. Otherwise, enroute.
|
Not much difference in setting up a cluster for z/OS. Using HOSTNAME works with z/OS like it does with UNIX - as long as HOSTNAME is discoverable- meaning that TCPIP has been configured in z/OS Unix System Services.
If you can successfully TCP PING the AIX HOSTNAME from z/OS, that means that your TCP/IP is configured correctly.
The issue here is that merely altering the CONNAME attribute on an active channel contributes a 2nd channel definition to the cluster while the first definition is still active. _________________ I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
Back to top |
|
 |
BKrunatrace |
Posted: Thu Oct 08, 2009 11:20 am Post subject: |
|
|
Newbie
Joined: 08 Oct 2009 Posts: 2
|
Think the problem is with the way MQ z/OS accesses the TCP/IP stack RESOLVER. You said that you could PING by name and establish MQ connection by IP and not name.
- I assume you issued the TCP Ping via TSO.
- and I assume you placed your hostname in a local z/OS host table
If that's the case, your TSO PING hits the Resolver each time.
MQ only reads it once at JOB startup.
Issue a RESOLVER REFRESH for the Z lpar/stack associated with your MQ Job/Channel. Syntax should be "F RESOLVER,REFRESH". This will force MQ to obtain the latest contents of the RESOLVER. For ex., if you placed your hostname in a Local Host Table, it will be seen. Alternatively you could restart your MQ JOB and take an outage....ouch!
This should fix your problem.
MQ uses a different instruction set than TSO to interface with the RESOLVER.
BK |
|
Back to top |
|
 |
BKrunatrace |
Posted: Thu Oct 08, 2009 11:29 am Post subject: |
|
|
Newbie
Joined: 08 Oct 2009 Posts: 2
|
BTW, that's also why it worked after you restarted your Q managers (you're getting the latest view of the RESOLVER).
A RESOLVER REFRESH after you add new names will prevent you from having to bounce your Q managers, atleast until IBM changes the way they interface with the z TCP/IP Resolver
BK |
|
Back to top |
|
 |
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|
|
|