Re: Oracle 8 and memory on NT --reboot?

I have customers running Oracle on NT that have to reboot daily and some that can go a month with out a reboot. Why? I have not been able to figure that one out. I even have one site with two different NT machines with the same hardware (major vendor), same rev of Oracle, same rev of the OS (NT 4 sp 4); neither one is a PDC or a BDC or a print server or a Web server and same basic database structure and similar numbers of users and one machine is slow and needs rebooting to speed it up (no evidence of memory leaks - not swapping and plenty of free ram) and the other is fine. It is probably some subtle registry setting somewhere that is killing one of the machines but not the other. (binaries are the same)

That is not a refuation of your statements, but in the NT world inconsistancy is the one constant things it seems.

Jim

dperez_at_juno_nospam.com wrote in message <36a28ff4.174402751_at_news.cwix.com>...
>On Sun, 17 Jan 1999 16:00:37 -0600, charlesm1NOSPAM_at_rocketmail.com (Charles
M)
>wrote:
>
>>In article <369ab225.80861723_at_news.cwix.com>, dperez_at_juno_nospam.com
>>says...
>>> OK, things are shutting down clean and starting UP clean..........
>>>
>>> BUT HOW DO I GET THIS POS TO NOT NEED TO BE SHUT DOWN ALL THE TIME?
>>> Oracle's had how long to figure out how to build development tools
without
>>> constant memory leaks? Sheesh!
>>>
>>
>>But is it Oracle or NT that's the problem here? Peruse the NT vs UNIX
>>posts and you'll notice that you don't see this problem with Oracle on
>>UNIX, but you do on NT. It may be that Oracle is as well written as it
>>can be on NT and that it's the OS that is the culprit.
>
>Um... Not to get into a pissing contest here, but this sounds like some of
the
>conversations I've had with Oracle support over the years...... THEY
insist
>that every problem is Microsoft's, and we always insisted that its ORACLE'S
>product, period, that is experiencing the product and should thus be
created
>with sufficient quality so as not to crash and burn even IF the operating
system
>isn't perfect! I believed, my coworkers believed, the DBAs believed, and
we ALL
>STILL BELIEVE that blaming everything on the operating system is a
convenient
>way of deflecting blame for shoddy engineering and utterly piss-poor
software
>quality control.
>
>So, while it is "nice" that Oracle dumped their product on every damn
operating
>system in existence, it would be even BETTER if they'd optimize the damn
thing
>for those operating systems so it wouldn't require CONSTANT nursing and
>babysitting.
Received on Sun Jan 17 1999 - 22:37:39 CST