Your post could have been written by me !
We had almost the problem on using IBM xSeries and FastT storage using
Windows 2000 server.
We started out on 9.0.1 patching it up to 9.1.0.4 - upgrading to 9.2.0.2.
Oracle blaming IBM, IBM blaming Oracle - firmware upgrades, endless
tar's, etc.
Our RAC hang every 3 to 4 days - bouncing one node caused the other one
to hang, split brains, bringin a node to 100% CPU and it crashed etc.
We've had a test setup on Compaq servers - experienced the same problem.
We had a test setup with HP wintel servers and SCSI disks - same problem.
After one year of terror - this was supposed to be a HA setup - we just
abandonned RAC.
I still have to see a stable RAC config on Wintel - it's a joke ! We made a
fail-over cluster
from our RAC - deinstalled the RAC option, maintained the raw devices, kept
one node
as production and we can fail-over to the other node by using custom
scripts.
The only platforms I still dare using RAC on is IBM AIX HACMP and Tru64
cluster.
"Steve Teeters" <teeterss_at_saic.com> wrote in message
news:fd5272ae.0306190728.4382f2d6_at_posting.google.com...
> We have a production installation of Oracle RAC (9.0.1.4.1). Running
> on Windows Advanced Server, w/2 nodes. Using Compaq/HP DL580 servers.
>
> We have been experiencing a problem where both instances will 'hang',
> meaning that all processing stops, and no users can log onto Oracle
> (not even SYS).
> We've had Oracle support helping us with this off and on, but with no
> success. We get absolutely no error messages in any of the logs;
> Oracle simply stops responding. The only way to get it back is to
> bounce both instances.
>
> Oracle support suggested we run a 'hang analyze' script (which they
> supplied), but we can't run ANYTHING when the hang occurs, so this
> script is useless.
>
> To get around this, we have been running with only one instance up....
> obviously getting NO benefit from having RAC at all. If we start both
> instances, we will get the hang problem within 2 or 3 days,
> guaranteed.
>
> Question: Is ANYONE out there having production success with RAC on
> Windows? Oracle is at a loss as to why this is happening, so I'm
> wondering if there's some help out there? Has anyone seen this 'hang'
> problem with RAC?
>
> Thanks,
> Steve
Received on Thu Jun 19 2003 - 13:18:28 CDT
