My Friday morning post somehow got lost, but ... Yes, I most definitely want to know if someone in the system is not playing by the rules documented in the Extended Assembler Guide. (It is documented that shared storage above the abr doesn't count against memlimit). My complaint is not that DB2 V8 *uses* the storage above the bar and I applaud the performance benefits that will hopefully give us. My complaint is that any bug in that new code for using storage above the bar can lead to a very hard outage called wait03C as opposed to 'just' an abend04E. And believe me, if the German stock exchange stops because of that and a limit that I cannot protect against - we will be in the media immediately not to mention having to write tons of reports to the CEO and the major German banks.
>A DB2 sub-system can't use 4GB for bufferspace, unless it gets defined >that way. Wrong. This summer we just had a bug in DB2 storage management that caused us AUX storage shortages (and that was 'old' code). Fortuntately, those were in non-priviledged address spaces, so they just 'stopped' until the shortage was relieved. With DB2 as authorized applications, we won't be so lucky. >If DB2 were to "use all that memory" I would need a large portion of my >entire DASD farm dedicated to page data sets. Isn't the max real >supported by z/OS still 128G? What's the point of hard coding 4T? Thanks for mentioning that, Mark. This is my point exactly, too. I get forced to provide DASD for the possibility of a runaway DB2 that I cannot control any other way. And our DBA's were sitting in when we decided on the limit for DB2, and we would provide a multiple of that value via USI, anyway. They would not go and define things bigger than we can handle. >>However, I do agree that DB2 should play by >>the rules and not override a system limit. And Teds disagreement aside - if a system wait states, the SYSPROG knew better when s/he wanted to impose a limit but couldn't. >Exactly. Nothing (that I know of) in the system (including *MASTER*) >has ever bypassed IEFUSI before. CA-OPS overwrites the LDA limits above the line. I got into a row with Asher Arembrand (Hi Asher, are you still lurking?) on that about 5 years back. Check the archives. But I also don't know any *IBM* product that would overwrite a system specified default with some arbitrary value. >Hopefully, DB2 uses SYSEVENT STGTEST or similar service to help it >intelligently use large virtual storage. Haven't traced that. But given IBMs conspicious silence on my questions (see my first post), my guss is that there is no 'official'/'IBM internal interface' to bypass IEFUSI. Ed, you yourself have proposed how to bypass USI in a prior post (and you even set RAXLVAUTH). The presence of that constant makes me believe that we'll see more of that, because 'IBM knows better'. So my main question again (maybe someone has an idea): Other than turning on VSTOR for DB2 private (which I rather wouldn't do because of the performance impact), is there any way to find out how much storage above the bar DB2's DBM1 address space is using? I want to catch that before we hit anything threatening the paging config by doing controlled IPLs should the need arise. Best regards, Barbara -- 10 GB Mailbox, 100 FreeSMS/Monat http://www.gmx.net/de/go/topmail +++ GMX - die erste Adresse für Mail, Message, More +++ ---------------------------------------------------------------------- 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
