In the 90's I was responsible for a product (Outbound, now ASG-Outbound Express, in case anyone remembers or still uses it) that used VTAM in batch. We would issue an OPNDST for an LU. If the OPNDST did not complete in a greater-than-reasonable amount of time, we needed to be able to cancel it and move on. (The problem only occurred in unusual circumstances; if the LU was in a "good" but disconnected state VTAM failed the request right away.) VTAM offered no way for an application to cancel a "hung" OPNDST. The OPNDST control blocks, including the ECB, were in GETMAIN storage that our logic freed and/or re-used. If the OPNDST ever completed (say the LU came on line later but while the jobstep was still running) or when we eventually closed the ACB, VTAM would try to post the ECB, get an ABEND, take a system dump, make our customers unhappy, etc., etc.
IBM eventually agreed that VTAM was missing a function, took an APAR, and implemented CLSDST OPTCD=TERMQ to provide the "cancel a hung OPNDST" function. It was a big help. It worked about 90% of the time. (The fix fell out of the next release of VTAM, but IBM dutifully put it back again. We got to know VTAM developers on a first name basis.) Somehow though, only 10% as many system dumps still didn't make the customers happy. We ended up putting the RPL and ECB for OPNDSTs in their own piece of GETMAIN storage (separate from our other session-related dynamic storage) in a unique sub-pool. Then, if the OPNDST did not complete one way or another, we simply abandoned this small bit of storage (300 bytes?) and left it allocated - in effect, introducing an intentional memory leak. This solved the problem, and did not create a problem of its own with storage leaks, because the problem only happened once or twice in a given batch jobstep, so the total memory leakage was only several hundred bytes - although in theory it could have been much higher. Perhaps you could consider a similar approach. Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of thomas zupar Sent: Wednesday, September 21, 2005 10:19 PM To: [email protected] Subject: Re: Cancel a VTAM post We have a situaton where the thread is being killed,so when VTAM gets back to post ,it finds the ecb wrong and hits with s0a8,thats why I am want to cancel the post. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
