a couple of HELP questions

[Jeff Manley]

Can anyone give me some insight into the following couple of issues?

[pdl :PDL :ERR /0:FROM_USTRANSCRIPTION_23:04/27/2009 00:06:37] read failed: Unknown error

[pdl :PDL :ERR /0:FROM_USTRANSCRIPTION_23:04/27/2009 00:06:37] read returned error 0 (No error)

[pdl :PDL :ERR /0:FROM_USTRANSCRIPTION_23:04/27/2009 00:06:37] PDL signaled exception: code 1, msg device error (remote side probably shut down)

[tps :tps :ERR /0:TO_USTRANSCRIPTION_23:04/25/2009 00:29:47] ‘KILL ‘ (returned by ‘check_ack ‘) does not match { }

[tps :tps :ERR /0:TO_LAB_MFN_22:04/16/2009 09:30:37] ‘KILL ‘ (returned by ‘check_ack ‘) does not match { }

Topic

[Max Drown (Infor)]

01) There is no connection to the sending/recieving application. I.e., the other side is down.

02) and 03)  Instead of using check_ack, try using hcitpsmsgkill. There is no need to use check_ack unless you want to do something different for each type of ACK/NACL. 99% of the time you’ll just want to kill the ACKs. If you truly need to use check_ack, I believe the argument list should look something like this: {{RETRIES 5} {SLEEP 60}}.

-- Max Drown (Infor)

[Jim Kosloskey]

Jeff,

For the second situation: The proc ‘check_ack’ has not provided a message handle for the message action (KILL being the action).

I would continue to use check_ack (or something like it) if I were you if you are pre Cloverleaf(R) 5.6 or are using the set of procs called recovery_xx (33, 56, etc.) as that proc also typically gets rid of the ob_saved_msg message. tpsmsgkill will get rid of the Reply. If you do not take care of the saved outbounf message, you are likely to get a memory leak.

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

[Russ Ross]

With regards to your error seen below

[tps :tps :ERR /0:TO_USTRANSCRIPTION_23:04/25/2009 00:29:47] ‘KILL ‘ (returned by ‘check_ack ‘) does not match { }

I use to experience this more frequently long ago and suspected it was a timing issue were the resend proc grabs control away from the check_ack procs before it finished.

I was able to almost completely illiminate the problem by increasing the amout of time between resends to higher time values like 60 seconds in most cases but as high as 5 minutes in really bad cases.

I also found including the msg kill function inside the check_ack also helped instead of having hcitpsmsgkill stacked behind check_ack.

Russ Ross
RussRoss318@gmail.com

[Mike Wilson]

Russ,

 

Where were you setting the resend value?  Thanks, I am seeing this on almost all of my threads.

Mike

[Russ Ross]

Yes the resend time out value for how long to wait before Cloverleaf attempts to resend a message that has yet to receive an ACK back from the foriegn system is the vaule that I increased until the problem mostly went away.

Russ Ross
RussRoss318@gmail.com

[Mike Wilson]

On the OUTBOUND tab in the INBOUND replies timeout portion on the bottom?[/img]

[Russ Ross]

Yes that is correct (see attached screen shot).

Also, if you set Timeout to -1 it will wait forever for an ACK and never resend.

Russ Ross
RussRoss318@gmail.com

[Mike Wilson]

Im already at 120, I will bump it up to 240 and see if that helps.  Thanks a lot!

[Author]

Replies