We have 2003 server and cluster running on 2 nodes. Each node has 4 GB RAM. Each node has one instance of SQL Server 2000 ent.

Recently we installed the /3GB switch and enabled AWE. We set max memory to 3.5 GB on each node.

Memory on one node (actually the less utilized node) has achieved 3.33 GB as shown in perfmon/memory manager/total server memory.

However, the other node (the more utilized node) has only achieved 2.7 GB.

The changes were made this weekend.

Any ideas why the one node is not using more memory?

Thanks for any suggestions.

You do not have enough memory free to stack all the instances on one node
should you have a failure. I would remove AWE (you don't need it anyway
with 4GB or RAM or less). /3GB may not be a good idea either in your
situation. I would set up each SQL instance with around 1.5 to 1.7GB of RAM
so that you have room for both on a single node during a failure.

Excerpt from Article 254321
-----------
Memory Allocation (All Versions)
When you allocate memory for clustered SQL Server servers, make sure that
the summed value of the maximum server memory settings for all the instances
of SQL Server plus any other cluster resource and local application
requirments is less than the lowest amount of physical RAM available on any
of the servers in the failover cluster.
-------------
--
Geoff N. Hiten
Microsoft SQL Server MVP
Senior Database Administrator
Careerbuilder.com



"Linda" <anonymous (AT) discussions (DOT) microsoft.com> wrote

I didn't set min memory. In the event of a failover, where both are running on one node, would they not just get as much as available

e.g: Instance 1 on Node A fails over to Node B. So both Instance 1 and Instance 2 are running on Node B. Prior to the fail-over, Instance 2 had 3.3GB. Now Instance 1 tries to start up. There is only 4GB total for the instances and the op sys. Wouldn't Instance 2 retain what it had, and Instance 2 try to grab what's remaining

If I want each instance to have 3.5 GB, then the total RAM on each node should perhaps be 3.5+3.5+.5+.5=8 GB

That would allow the fail-over total to accommodate both instances at the 3.5 level

Thanks

Answes Inline
"Linda" <anonymous (AT) discussions (DOT) microsoft.com> wrote

Memory. MIN memory is ignored.

3.3GB. Now Instance 1 tries to start up. There is only 4GB total for the
instances and the op sys. Wouldn't Instance 2 retain what it had, and
Instance 2 try to grab what's remaining?
will have overcommitted physical RAM and will page heavily. Performance
will be very poor and your server may crash.



Again, you don't need AWE with only 4GB physical memory on a server. You
will need it

In message <uTewGus3DHA.2168 (AT) TK2MSFTNGP12 (DOT) phx.gbl>, Geoff N. Hiten
<SRDBA (AT) Careerbuilder (DOT) com> writes
Geoff,

What actually happens if you don't follow this rule? Does anything die?

Won't the existing instance just get asked by the OS for some memory
back, give it up nicely and everything runs a bit slower for a while.

In many cases I suspect the idea of multiple instances giving maximum
use of the hardware, but with a performance degradation during a
failover situation would seem quite appealing as opposed to extra cost
and redundant hardware.

--
Darren Green (SQL Server MVP)
DTS - http://www.sqldts.com

PASS - the definitive, global community for SQL Server professionals
http://www.sqlpass.org

Answers Inline
"Darren Green" <darren.green (AT) reply-to-newsgroup-sqldts (DOT) com> wrote


actually get virtual memory, your system will start paging like crazy and
will eventually suffer latch wait timeouts when it can't get pages into and
out of the paging file fast enough. At that point, SQL is toast and will
require restarting. Either that or the failed instance won't come up at
all.




--
Geoff N. Hiten
Microsoft SQL Server MVP
Senior Database Administrator
Careerbuilder.com