Bug 116733

Summary: kernel hangs or crashes when hugetbl is on and Oracle 10g uses it
Product: Red Hat Enterprise Linux 3 Reporter: Gunther Schadow <ebiz>
Component: kernelAssignee: Larry Woodman <lwoodman>
Status: CLOSED NOTABUG QA Contact: Brian Brock <bbrock>
Severity: high Docs Contact:
Priority: high    
Version: 3.0CC: petrides, riel
Target Milestone: ---   
Target Release: ---   
Hardware: i686   
OS: Linux   
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2004-10-20 21:21:37 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Description Gunther Schadow 2004-02-24 19:04:48 UTC
Description of problem:

On a system with 20GB ram running the RHEL3 hugemem kernel
and hugetbl_max set to 8192 (which strangely enough 
/proc/meminfo reports as being just 4096 pages, and it 
looks like meminfo is right) when I run Oracle 10g using
lots of memory the kernel suddenly panics or just simply 
locks up.

Version-Release number of selected component (if applicable):

How reproducible:

Happens every other time I start Oracle (with a configuration
that really uses lots of memory)

Steps to Reproduce:
1. Install Oracle 10g
2. modify oracle to lower-mapped-base specs (optional)
3. modify oracle to enable aio (optional)
4. echo 8192 > /proc/sys/kern/hugetbl_max
5. specify sga_max_size = 3500M
6. specify db_buffer_cache = 2000M

--> of course I don't expect you to do all this, just to 
    debug a kernel error. I would make a program that 
    actually allocates lots of hugetbl shared memory
    and uses it heavily.

Actual results:

system crashes/hangs

Expected results:

system runs happily :-)

Additional info:

Comment 1 Suzanne Hillman 2004-02-26 22:21:29 UTC
Please include information about the hardware being used, as well as
information about whether or not it happens on multiple machines.

Comment 2 Gunther Schadow 2004-03-04 01:51:13 UTC
Hardware is a ProLiant DL780 (?) or so, it's got 4 CPUs.
I can't tell whether it happens on multiple machines. What
I do know is that when it actually panics and I get to
see the dump output, it does mumble something about hugetbl

Comment 3 Larry Woodman 2004-04-05 15:50:33 UTC
Can you please get several "AltSysrq M" outputs when the system is
hung while running 10g with hugltlb pages?  I cant reproduce this
problem locally.

Thanks, Larry Woodman

Comment 4 Larry Woodman 2004-04-20 15:04:17 UTC
Please try to reproduce this problem with the latest RHEL3-U2 kernel
(2.4.21-14.EL).  We found and fixed a problem which sounds exactly
like this one in that kernel.

Larry Woodman

Comment 5 Larry Woodman 2004-08-02 21:07:49 UTC
Gunther, whats the status of this bug?  I havent seen any update in
the last 3 months.


Comment 6 Suzanne Hillman 2004-10-20 21:21:37 UTC
Closing due to lack of activity. If there is additional information, please
reopen and provide the information.