Re: 9i RAC or 10g RAC ?

From: Matthew Zito <mzito_at_gridapp.com>
Date: Tue, 11 May 2004 07:57:49 -0400
Message-Id: <6C7FE68A-A342-11D8-93AF-000393D3B578@gridapp.com>

In Linux RAC, the hangcheck-timer's job is to reboot the machine should a hung node every come back. The flow of the kernel module looks something like:

-Set an alarm to be woken up in X seconds
-Go to sleep
-I've been woken up - what time is it?
-If Y amount of time more than the X seconds has passed since the last

time I woke up, the machine was hung.
-If the machine hung, crash the system.
-If not, then set a timer to be woken up in X seconds again

So, if a Linux server truly hung, the hangcheck-timer won't reboot it unless it suddenly fixes itself. The point is to keep a server from hanging, another node starting recovery, the first server suddenly coming back and corrupting data.

Thanks,
Matt

--
Matthew Zito
GridApp Systems
Email: mzito_at_gridapp.com
Cell: 646-220-3551
Phone: 212-358-8211 x 359
http://www.gridapp.com


On May 10, 2004, at 10:54 PM, zhu chao wrote:



> During my 2 years' experience running both linux and solaris,

> single instance oracle on linux never crashed. Single instance on 

> solaris

> crashed 1 time.

>

> 3. Unknown reason, one node hang with one session reporting 7445. Whole

> machine hangs. Hangcheck does not reboot the instance.

>

>     If for high avaliable, RAC is indeed not a very good solution, 

> unless

> you can always wait to 8.1.7.4 or 9.2.0.5 etc, and you application is 

> really

> stable and designed for RAC. RAC is more easy to crash indeed.

>

> Regards

> Zhu Chao.

>


----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to:  oracle-l-request_at_freelists.org
put 'unsubscribe' in the subject line.
--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Received on Tue May 11 2004 - 06:54:20 CDT