Cloverleaf(R) 5.6 Auto-resend assistance

[Jim Kosloskey]

All,

I am trying to experiment with the new Auto Resend feature in Cloverleaf (R) 5.6 (Rev 2).

I have an Outbound thread. The Outbound Tab has Timeout Handling set for ‘Resend OB message’. Timeout is set to 10 seconds.

I do not have anything in Send OK Procs.

I do have a TPS Inbound Reply proc but that never gets executed as it should not given the following.

I am sending locahost to an Inbound thread tha is not configured to send a reply. Everything gets connected properly and I resend one Outbound mesage from the Outbound thread.

I would expect to see the Outbound thread resend the same message every 10 seconds until I stop it.

That is not occurring and here is what I have in the Log:

[pd  :pdtd:INFO/1:       conn_4:08/21/2009 14:10:20] Timed out while awaiting replies on thread. While no reserved OB Message to send

[pti :even:DBUG/0:       conn_4:08/21/2009 14:10:20] Unregistering TIMER () event 0x212055c8 for tid 2

[pti :even:DBUG/0:       conn_4:08/21/2009 14:10:20] evUnregister TIMER event 0x212055C8 for tid 2

[pd  :pdtd:INFO/0:       conn_4:08/21/2009 14:10:20] Shutting down await reply mode

[dbi :rlog:DBUG/3:       conn_4:08/21/2009 14:10:20] [0.0.16745] Looking for mid in recovery db

[dbi :dbi :DBUG/1:       conn_4:08/21/2009 14:10:20] (0) ‘cl_keyfindlock: About to do d_keylock’

[dbi :dbi :DBUG/1:       conn_4:08/21/2009 14:10:20] (1000) ‘cl_keyfindlock: About to do d_keyfind, keyval:’

[pd  :pdtd:INFO/3:       conn_4:08/21/2009 14:10:20] [0.0.16745] Unable to delete reserved OB message from recovery database!

msg: 0x3000e03c

   msgType           : DATA

   msgClass          : PROTOCOL

   msgState          : OB reserved for IB Reply (16)

   msgPriority       : 5120

It appears to me the timeout has expired and Cloverleaf does not think it has a reserved message to resend. Yet it later complains it cannot delete the reserved message from the DB. Indeed the reserved message is there (I only showed the beginning of the Message Dump).

In any case what follows the dump of the message is just the normal noise of Cloverleaf with nothing to do.

I must have something not configured correctly but I have looked at this a few times and do not see there is anything additionally I need to configure and what I have configured seems to be correct.

Can anybody give me a suggestion as to what I am doing wrong?

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

Topic

[Keith McLeod]

I am only guessing here but do you have retries for outbound data set to a value other than -1.  Maybe the 10 second timer goes here.  Let me know, I have not done this yet…

[Jim Kosloskey]

Keith,

No, I left that (Retries in Outbound Panel) set to the default (-1).

Besides, that is ofr protocol not message as I understand it.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Bob Richardson]

Jim and company,

Normally on Sundays I don’t work (trying to respect the Sabbath) but being on call this weekend I found your post while checking up on stuff (a favorite Minnesotan word).  Anyway, hopefully this information will prove useful.

We discovered about a year ago in evaluating CIS5.6 that the auto reply/resend only works with the PDL TCP/IP protocol configuration and nothing else.  And in the TPS IB Reply context you need to plug in a 5.6 updated version of hcitpsmsgkill which interprets the OBMSGID key for the State 16.  We got our version from Charlie Bursell and placed it in our

master site (which is soft linked to by our other sites).  We did request an enhancement to include (what we call) native TCP/IP protocol in this functionality but haven’t heard back since.

If you want to do reply/resend otherwise, you need to implement the same model as the old 33_Recovery setup;  the new 56_Recovery procs are mostly noops anyway and only its resend procedure has any real message resend capability.

Let me know if this was helpful and addressed your problems.

Try to enjoy the rest of the day!

[Jim Kosloskey]

Bob,

Thanks for that info it definitely helps.

I will try it with PDL/TCP. I am just trying to get a feel for the process so it does not matter to me at this time that is resticted for protocol.

I agree however, that the facility ought to be available of any protocol that has the potential for acknowledgments.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Tom Patton]

Jim,

I just set this up as a PDL/TCP with timeout set to 30 secs and with the 5.6 tpsmsgkill in place and it worked as I hoped / wanted.

I needed to do this b/c we’re throwing away msgs after 3 tries with the current setup on some threads.

What I did was remove the reply proc from the localhost receive side and the engine kept retrying every 30 (my setting) seconds.  When I added the reply proc on the receive and cycled the process the msgs all got sent and cleaned up.

So the net of this is to get the tpsmsgkill_56 and put it in the Outbound tps IR reply procs and it seems to work.  

Thanks for the question b/c I was halfheartedly trying to figure out how this might work and now I know.

[James Cobane]

All,

We have the majority of our threads configured to ‘Resend OB’ on a timeout if a reply is not received; generally at 60.

[Author]

Replies