Bruce My first reaction is that my objective with the document which I quoted was to "touch all the bases" involved in having two instances of Communications Server IP running on one OS/390 - as it was called that week (in 2001). This was to guide whomever of the system programmers was going to solve a particular problem of consolidating the applications of more than one business unit onto one OS/390 image on what needed looking into. It was 5 years ago now and I don't recall whether or not I paid much attention to the actual values. I may have picked the defaults or maxima that applied at the time and assumed that the eventual systems programmer would be paying more attention to the numbers when the production implementation was put in place.
You'll note I put particular attention on which of perhaps a plethora of parameters need to be reviewed, I realise that 5 years later this doesn't look so clever but, assuming the work was to be done with the then current level of the software, this seemed to be a helpful point to cover. Now I'll look into your precise question. I'd prepared my answer to this point before performing some research. In the meantime Peter has provided his opinion - and I agree - it doesn't make sense to me either. Your quote implies a relationship between INADDRPORT and MAXSOCKETS - as you indicated. If there were to be any relationship here I could imagine one between INADDR*COUNT* and MAXSOCKETS. My rather basic exposure to sockets programming leads me to expect that each "client"-type socket will use one ephemeral port. Since there will be a number of "server"-type sockets, this will tend to mean there will be a need for more sockets than ephemeral ports. Thus MAXSOCKETS should be > than INADDRCOUNT. Probably something like this was suggested to the manual author verbally during a manual review meeting by a developer and the author got confused and made a mistake in his notes of the meeting. Chris Mason ----- Original Message ----- From: "Bruce Hewson" <[EMAIL PROTECTED]> Newsgroups: bit.listserv.ibm-main To: <IBM-MAIN@BAMA.UA.EDU> Sent: Tuesday, 10 October, 2006 9:23 AM Subject: Re: TCP/IP disabled if BPXPRMxx holds AF_UNIX defs > Hello Chris, > > On Mon, 9 Oct 2006 11:41:46 +0200, Chris Mason <[EMAIL PROTECTED]> > wrote: > > <snip> > > > >BPXPRMT1 Extract > > > <snip> > > INADDRANYPORT(61000) > > INADDRANYCOUNT(4000) > > MAXSOCKETS(64498) > <snip> > <snip> > >INADDRANYPORT > > > >The INADDRANYPORT parameter specifies the first port number in the range of > >ports which are reserved for dynamic assignment of port numbers when no > port > >number and no IP address are specified in a bind() request. It is used in > >conjunction with the following INADDRANYCOUNT parameter. > <snip> > > > Please explain your choice of INADDRANYPORT < MAXSOCKETS in respect to this > statement from the manual. > > 3 INADDRANYPORT and INADDRANYCOUNT specify the first ephemeral port number > and the range of ports for z/OS UNIX CINET usage. The starting port number > should be set at least as high as the value for MAXSOCKETS. > > taken from: > z/OS Communications Server > IP Configuration Guide > Version 1 Release 6 > Document Number SC31-8775-06 > > Regards > Bruce Hewson ---------------------------------------------------------------------- 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
