Re: active-pasive cluster in linux
- From: John Thompson <jhthomp@xxxxxxxxx>
- To: przemolicc@xxxxxxxxx
- Date: Tue, 23 Feb 2010 11:18:13 -0600
Our sysadmins insisted on doing fschk's on a server crash. On a 3tb system,
it takes awhile.
2010/2/23 <przemolicc@xxxxxxxxx>
> On Tue, Feb 23, 2010 at 09:47:54AM -0600, John Thompson wrote:
> > We've got mostly 2 node (but also have a 6 and 4 node) clusters setup
> using
> > CRS. Oracle binaries on OCFS2. Storage on ASM. Multiple database in
> > active/passive mode on each cluster. This is all on Redhat Linux on
> > commodity hardware. Prior to that, we were on Solaris using Veritas file
> > systems. If a database crashed in the Solaris/Veritas environment, it
> would
> > take 45 mins for that server to reboot and mount all the file systems for
> > all the databases. With out current environment, there's no need to
> > mount/fschk file systems upon reboot so our failover time is usually less
> > than 5 mins. We also reduced the number of licenses needed by going from
> > the Solaris full license price to the 1/4 priced AMD's. Uptime has also
> > increased dramatically as the Sun 2900's and 6800's were not as fault
> > tolerant as advertised.
>
> VxFS is journaled filesystem and in case of unplanned reboot it takes
> really a few seconds (up to 5-10 - depends on filesystem size) to check
> its consistency.
> There is (usually) no need to run fsck on VxFS.
>
>
> Regards
> Przemyslaw Bak (przemol)
> --
> http://przemol.blogspot.com/
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> ----------------------------------------------------------------------
> Wygraj iPhona >>> http://link.interia.pl/f25d5
>
> --
> //www.freelists.org/webpage/oracle-l
>
>
>
Other related posts:
