Hi Morag,
thanks for the clarification. Too bad that the manuals on this subject is
not very clear. I'll check that setting with our mainframe programmer.
best regards,
Benjamin F. Zhou
Messaging & Integration
Enterprise Applicatin Integration
Mercedes-Benz USA
x.2474
Morag Hughson
<[EMAIL PROTECTED] To: [EMAIL PROTECTED]
COM> cc:
Sent by: Subject: MQ Channel KeepAlive not
working on os/390
MQSeries List
<[EMAIL PROTECTED]
en.AC.AT>
11/12/2003 03:47
PM
Please respond
to MQSeries List
I agree with you that the Intercomms guide and MQSC book could both do with
an addition something along the lines of that which is in the System Setup
Guide, thus:-
TCPKEEP
Specifies whether the TCP KEEPALIVE facility, as specified by the
KEEPALIVEOPTIONS statement in the TCP profile configuration data set,
is to be used. If it is used, the interval is specified by the
channel attribute KeepAliveInterval (KAINT).
TCPIP Keepalive is a TCPIP feature and requires the overall TCPIP keepalive
to be switched on. What it does not require you to do is change the overall
system timeout value. That can be left at 2 hours, or whatever it is on
your system. It must be switched on though.
I will look into getting the other books updated with this information
Cheers
Morag
Morag Hughson (Miss)
WebSphere MQ for z/OS Development
Telephone: +44 (0) 1962 816900
Internet: [EMAIL PROTECTED]
------------------------------------------------------
HI,
as Morag suggested below and in his article. I set the KEEPALIVE to YES in
the CSQ6CHIP macro for the queue manager, set KEEP ALIVE INTERVAL to 120
seconds, HBINT to 300 seconds so it doesn't interfere with the test. I did
the same on the sending side of the channel on AIX, and set DISCINT to 6000
seconds; I started the channel by sending a message to a remote queue, saw
the channel running, then pulled the network cable from the AIX box.
However, after over 60 minutes, the rcvr channel on os/390 is still in RUN
status. Apparently, not even the heartbeat interval is functioning as
described in Horag's document page 19-20. It was my understanding before
that with network interrupted, HBINT and DISCINT won't work. But Horag's
article suggested otherwise, and said heartbeat interval is used to detect
network failure, which is more plausible than my initial understanding.
The manual also implies that on os/390 (z/os), TCPIP KeepAlive KAINT is
mqseries specific so there is no need to reconfigure KeepAlive for the
system's overall TCPIP settting. Am I wrong on this?
Has anyone tested this attribute on os/390?
best regards,
Benjamin F. Zhou
Messaging & Integration
Enterprise Applicatin Integration
Mercedes-Benz USA
x.2474
Morag Hughson
<[EMAIL PROTECTED] To:
[EMAIL PROTECTED]
COM> cc:
Sent by: Subject: Re: MQ Channel
KeepAlive not working on MVS
MQSeries List
<[EMAIL PROTECTED]
en.AC.AT>
10/30/2003 06:06
AM
Please respond
to MQSeries List
Ensure that you have TCPKEEP=YES set in your channel initiator parameters
(CSQXPARM built from CSQ6CHIP Macro) and also ensure that KeepAlive is set
on in your TCP/IP settings.
Quick plug for new SupportPac MA0C
http://www-3.ibm.com/software/integration/support/supportpacs/individual/md0c.html
See page 25-26 for keepalive details.
Also see the Intercommunication guide, Chapter 6. Channel Attributes
Cheers
Morag
Morag Hughson
WebSphere MQ for z/OS Development
Telephone: +44 (0) 1962 816900
Internet: [EMAIL PROTECTED]
------------------------------------------------------
Hi,
on MVS, Keep Alive Interval is a channel attribute. To test its
effectiveness, I set it to 30 seconds on a receiving MCA on mainframe,
start the channel, then unplug the network cable from the sending side (AIX
box), hoping the status of the receiving MCA on mainframe will become
inactive in about 30 seconds. But it didn't. I keep the cable off for 5
minutes, the receiving MCA is still in running status. (Note, to remove
the influence of HBINT and discint in the test, I set them both to very
large values)
However, after I plug the cable back, the sending channel goes into healthy
running state right away.
I tested the same on NT machines. there, after 30 seconds, the receiving
side becomes inactive and ready to be restarted. This is what I had
expected based on my knowledge on TCP/IP KeepAlive. But why is it different
on MVS?
Is anyone working on mainframe aware of this? Can you offer some
explanation?
thanks,
Benjamin F. Zhou
Messaging & Integration
Enterprise Applicatin Integration
Mercedes-Benz USA
x.2474
Instructions for managing your mailing list subscription are provided in
the Listserv General Users Guide available at http://www.lsoft.com
Archive: http://vm.akh-wien.ac.at/MQSeries.archive
Instructions for managing your mailing list subscription are provided in
the Listserv General Users Guide available at http://www.lsoft.com
Archive: http://vm.akh-wien.ac.at/MQSeries.archive
Instructions for managing your mailing list subscription are provided in
the Listserv General Users Guide available at http://www.lsoft.com
Archive: http://vm.akh-wien.ac.at/MQSeries.archive
Instructions for managing your mailing list subscription are provided in
the Listserv General Users Guide available at http://www.lsoft.com
Archive: http://vm.akh-wien.ac.at/MQSeries.archive
