I'm not sure if this applies to your situation, or not. I saw a similar timeout problem with HATS applications at my previous (now-defunct) shop. The problem was due to the fact that HATS did many, many resolve requests to the nameservers, which were located on Unix platforms. We had to run a cacheing nameservice on the mainframe (on the same LPAR) to resolve this.
-Bob Shade 253-657-8622 -----Original Message----- From: IBM Mainframe Discussion List [mailto:ibm-m...@bama.ua.edu] On Behalf Of Chris Mason Sent: Tuesday, April 27, 2010 8:09 AM To: IBM-MAIN@bama.ua.edu Subject: Re: HATS client timeout/performance issue Bob > ... (... and TCPIP) That's actually IBMTCP-L where matters concerning the IP component of z/OS Communications Server - and sometimes TCP/IP for VM - tend to attract the widest knowledgeable audience. Because of the importance of Enterprise Extender for SNA support, the SNA component often attracts suitable cognoscenti there also. I think the nearest I ever got to HATS (= Host Access Transformation Server) was "Host Publisher" back in 1999. I see there is a handy "red paper", "Host Access Transformation Server Concepts and Architecture", at which I am going to take a look. Meantime I'm switching to IBMTCP-L and covering NODELAYACKS - which, IMNSHO, is blameless! Chris Mason On Tue, 27 Apr 2010 07:53:39 -0400, Richards, Robert B. <robert.richa...@opm.gov> wrote: >Cross post (IBM-Main and TCPIP) > >We are experiencing an unusual situation and I am wondering if anyone else has seen this at their shop. As phrased by my boss: > >"We are currently experience a client timeout with HATS applications. I'm calling it a client timeout because the CICS session associated with the user is active but the user received a timeout message. This issue is only with HATS applications and cross over business application lines." > >My TCP/IP guy wants to implement NODELAYACKS for specific ports for problem resolution per the following IBM recommendation for TCP/IP WAS performance tuning: > >- Consider specifying NODELAYACKS for ports. This may improve throughput for more trivial transactions, but it does add more overhead for more complex transactions. Using this option will cause the acknowledgments (ACKs) to be sent back immediately to the client, rather than waiting for more data to accumulate. > >Because CICS still has the session as active, it holds on to the client's too long and, as you might expect, that has a bad ripple effect until CICS terminates the task at its timeout. > >Any takers? :) > >Bob ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
