T.R | Title | User | Personal Name | Date | Lines |
---|
194.1 | Contact Oracle, And Ask Them | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Fri Feb 14 1997 10:49 | 18 |
|
Hi Sally -- you really need to contact the Oracle folks directly.
OpenVMS does not have something called a `process heap size' that
can be increased.
Thus this isn't directly an OpenVMS question and the customer should
not be asking us -- if an Oracle product or package is outputting an
ambiguous message, a message with insufficient information to recover
from the problem, or a bogus error message, then Oracle needs to update
the message or the error messages and recovery text. And hopefully,
Oracle (support) also knows what this error message is trying to tell
the OpenVMS system manager to do...
Oracle Rdb stuffs some recovery text into SYS$HELP:*RDB*.*.
See HUMANE::SYS$PUBLIC:EASYNOTES.LIS -- there are a couple of
Oracle notes conferences listed there.
|
194.2 | heap ~= program's dynamic memory | GREGOR::OPP | | Fri Feb 14 1997 13:19 | 35 |
| One definition of "heap", as found at:
http://nsl-too.pa.dec.com/doc/OSF1_alpha/V4.0/AA-Q2DPC-TKT1_html/thrd0059.html
is as follows ...
3.3.1.3 Heap
Dynamic memory is allocated by the program as a result of a call to some
memory management function (for example, the C language run-time function
malloc() or the OpenVMS common run-time function LIB$GET_VM).
Dynamic memory is referenced through pointer variables. Although the pointer
variables are scoped depending on their declaration, the dynamic memory
itself has no intrinsic scope or lifetime. It can be accessed from any
routine or thread that is given its address and will exist until explicitly
made free. In a language supporting automatic garbage collection, it will
exist until the run-time system detects that there are no references to it.
(If your language supports garbage collection, be sure the garbage collector
is thread-safe.)
Heap is usually appropriate to manage persistent context. For example, in a
thread-reentrant routine that is called multiple times to return a stream of
information (for example, to list all active connections to a server or to
return a list of users), using dynamic memory allows multiple contexts that
are independent of threads. Multiple threads may be able to share a given
context, or a single thread may have more than one context.
The scope of dynamic memory is anywhere a pointer containing the address can
be referenced. The lifetime is from allocation to deallocation.
Based upon this description, I'd try increasing the process' WSQUOTA
and WSEXTENT assuming the C programs are *not* using global pages as "heap".
Greg
|
194.3 | Page file quota? | VIRKE::GULLNAS | Olof Gulln�s, DTN 876-7997 | Sun Feb 16 1997 13:44 | 8 |
| .2:
Page file quota is more likely than wsquota or wsextent.
Might even be the VIRTUALPAGECNT sysgen parameter.
As .1 said, without more information about hat the message means,
we are just guessing...
Olof
|
194.4 | need more info from Oracle | MIASYS::SSAFFER | | Wed Feb 19 1997 09:48 | 20 |
|
Thank you for the suggestions. I called the customer to see what his
UAF params were already set to:
pgflquo 500000
wsquota 4000
wsextent 131000
These seemed sufficient, especially with an extent as large as
that, but just for kicks, I had him increase
wsquota to 20000, but it didn't make a difference.
I suggested that he call Oracle again and ask them to take another look
at the problem. These are Oracle sample programs that don't work and
they should be supported. I told the customer that unless Oracle could
narrow the problem down to an OpenVMS issue, that we really could not
continue to 'guess' at what the problem might be. I will continue to
monitor the customer's progress and will post the solution here when it
is resolved.
|
194.5 | ?pointer size?? | GREGOR::OPP | | Thu Feb 20 1997 07:03 | 9 |
| SWAG:
Is this version of Oracle using 64-bit pointers or 32-bit
pointers or a mix of both? This could make quite a difference
in trying to compile a sample application. Refer to "Digital
Technical Journal", Volume 8 Number 2, 1996 for three excellent
articles on OpenVMS for 64-bit addressable virtual memory.
Greg
|