[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference vaxaxp::vmsnotes

Title:	VAX and Alpha VMS
Notice:	This is a new VMSnotes, please read note 2.1
Moderator:	VAXAXP::BERNARDO

Created:	Wed Jan 22 1997
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	703
Total number of notes:	3722

526.0. "NPAGEVIR greater than 1GB?" by WBC::STASSE (Glenn Stasse, 339-7780) Thu Apr 24 1997 18:46

    Greetings all,
    
    A customer needs to set NPAGEVIR to a value greter than 1GB. I can set
    the value to ((1024*1024*1024)*2)-1 and the command will take it. I can
    set NPAGEVIR to %xFFFFFFFF and it shows up as -1 (presumably, the max)
    but either way, the system crashes instantly after typing 'continue'.
    I've been able to set NPAGEVIR as high as 1GB and the system works. I
    can set it to 1.5 GB and the system boots but the database says
    insufficient PTEs when we try to open it at NPAGEVIR > 1GB. When I used
    AUTOGEN to set min_npagevir=2147483648 I wound up with a value around
    308MB.
    
    Particulars: OpenVMS/AXP v7.1, OracleRDB 7, AS4100 w/ 4GB.
    
    I tried the reserved memory register but we don't know a way to make
    the database use the reserved section of memory that results.
    
    This is apparently the VLM methodology for RDB systems. The RDB guy
    says they've done 2 GB databases in memory a lot. We've missed
    something, clearly, but don't know what.
    
    Ideas, constructive suggestions, etc. most welcome
    Thanks...

T.R	Title	User	Personal Name	Date	Lines
526.1		UTRTSC::utojvdbu1.uto.dec.com::JurVanDerBurg	Change mode to Panic!	`Fri Apr 25 1997 04:02`	5
	You're trying to setup a database in nonpaged pool???? Are you sure that's what you need? Jur.
526.2	Check Rdb Documentation...	XDELTA::HOFFMAN	Steve, OpenVMS Engineering	`Fri Apr 25 1997 10:08`	21
	The page tables are extended automatically, assuming sufficient available physical memory exists. Non-paged pool does occupy part of the page tables, but it does not limit the size of the page tables -- the page tables are in a seperate part of 64-but memory. (PT-space) You'll need to check with the Oracle Rdb documentation and the Rdb support folks for information on setting up Rdb on V7.1 -- that you are trying to use non-paged pool for Rdb sounds rather strange. (I thought Rdb used S2 (or P2) space, not S0-space non-paged pool.) You may want to take the CLUE output from one of the crashes, and send it to CANASTA -- see 233.* here -- and get this crash entered into the CANASTA database, if it's not already there... (It would appear that you've filled all of S0 space with non-paged pool, and left no room for the rest of the operating system...) Also see the Oracle Rdb notes conference at ORAREP::NOMAHS::RDB_70.
526.3	Use the row cache, not bind buffers	WBC::STASSE	Glenn Stasse, 339-7780	`Mon Apr 28 1997 14:22`	17
	Well here's the answer for those who may follow.... The way to get RDB to use memory the way you'd expect, rather than out of non-paged pool, is to alter the table's row cache, i.e. add cache TABLE_NAME cache size is 1000000 rows row length is 152 bytes .... etc. The WRONG way to add memory (and the way to get data cached in pool) is to increase bind buffers in the database startup file. This has a host of effects I don't understand but you don't want to do it, trust me. Setting that number real high has the effect of caching the data twice, so I understand. Glenn