I know that RAC and Oracle 13 are not yet supported on OL 6.x, but the
installation goes without a hitch. The only additional thing that needs
to be installed is the compatibility version of the libcap library:


[oracle@rac1 ~]$ rpm -qa|grep compat-libcap
compat-libcap1-1.10-1.i686
[oracle@rac1 ~]$

Of course, the installer must be invoked with -IGNORESYSPREREQS and cluvy
at the end of the installation will fail because "this system doesn't
satisfy the requirements", but it will work without a hitch.


[oracle@rac1 ~]$ uname -a
Linux rac1 2.6.32-100.34.1.el6uek.i686 #1 SMP Wed May 25 17:28:36 EDT
2011 i686 i686 i386 GNU/Linux
[oracle@rac1 ~]$ cat /etc/oracle-release
Oracle Linux Server release 6.1
[oracle@rac1 ~]$

The installation finished without a hitch:
[oracle@rac1 ~]$ crsctl stat res -l
NAME=ora.LISTENER.lsnr
TYPE=ora.listener.type
CARDINALITY_ID=rac1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.LISTENER_SCAN1.lsnr
TYPE=ora.scan_listener.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.asm
TYPE=ora.asm.type
CARDINALITY_ID=rac1
DEGREE_ID=1
TARGET=OFFLINE
STATE=OFFLINE

NAME=ora.cvu
TYPE=ora.cvu.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.gsd
TYPE=ora.gsd.type
CARDINALITY_ID=rac1
DEGREE_ID=1
TARGET=OFFLINE
STATE=OFFLINE

NAME=ora.net1.network
TYPE=ora.network.type
CARDINALITY_ID=rac1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.oc4j
TYPE=ora.oc4j.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.ons
TYPE=ora.ons.type
CARDINALITY_ID=rac1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.rac.db
TYPE=ora.database.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.rac1.vip
TYPE=ora.cluster_vip_net1.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

NAME=ora.scan1.vip
TYPE=ora.scan_vip.type
CARDINALITY_ID=1
DEGREE_ID=1
TARGET=ONLINE
STATE=ONLINE on rac1

[oracle@rac1 ~]$ srvctl status database -d rac
Instance rac1 is running on node rac1


[oracle@rac1 ~]$ sqlplus scott/tiger

SQL*Plus: Release 11.2.0.3.0 Production on Tue Nov 22 20:24:02 2011

Copyright (c) 1982, 2011, Oracle. All rights reserved.


Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - Production
With the Partitioning, Real Application Clusters, OLAP, Data Mining
and Real Application Testing options

SQL>

Oracle will probably not support this, but it is worth doing. The
commands like iotop or atop with "-D" option now work and will show you
the right results. The "perf top" command still doesn't work properly, it
doesn't show me anything on OL 6.1
This version of RAC is installed using OCFS2. I have also verified the ASM
install, which works flawlessly but my machine is too small. The ASM
version was so slow that it was useless. Raw devices on OL 6.1 are
created the same way as on RH 5.7: using /etc/udev/rules.d/60-raw.rules.
The unintended consequence was that the protection of /dev/null was
changed to 0660 but once I fixed that by putting the corresponding line
in the 60-raw.rules, everything worked well.
My only objection to OL 6.1 is the fact that "perf top" doesn't work. I
am not on ULN, maybe there is a patch for that.
The fact that Red Hat is disabling OCFS2 in their kernels is sufficient
for me to recommend OL instead of RH for Oracle servers. The fact that OL
6.1 support comes with XFS, according to Mathias Hoys, is an additional
plus. OL 6.1 is a viable alternative to Red Hat.
I was even able to get RAC working on Fedora Core 15, with the
compatibility rpm mentioned above. Unfortunately, Fedora doesn't support
either raw devices or ASMLib, ASM is not possible. This was, of course,
just a test, nobody in the right mind would install RAC in the office. It
does, however, show one thing: RAC is now a decently polished product
which is easy to install. Installing 10g RAC required a lot of experience
and nerves of steel. It would normally fail if anything was wrong. With
11g, it is getting hard to have an unsuccessful installation, even if the
OS is totally wrong, like FC 15 is. RAC has matured.

--
http://mgogala.byethost5.com

On Wed, 23 Nov 2011 01:30:51 +0000, Mladen Gogala wrote:

oracle 11.2.0.3



--
http://mgogala.byethost5.com

On Nov 23, 2:30*am, Mladen Gogala <gogala.mla... (AT) gmail (DOT) com> wrote:

Regarding XFS, this comes from the OL6.1 release notes (http://
oss.oracle.com/el6/docs/RELEASE-NOTES-U1-en.html):

"The following Technology Preview features are currently not supported
under Oracle Linux 6 and may not be functionally complete:
....
XFS
....
These features are not suitable for production use. However, these
features are included to provide the feature with wider exposure."

However, in the OL6.0 release notes (http://oss.oracle.com/ol6/docs/
RELEASE-NOTES-GA-en.html), XFS is NOT included in the list of
"Technology Preview features"... So, does this mean that they removed
support for XFS between versions 6.0 and 6.1???

Matthias Hoys

On Wed, 23 Nov 2011 00:47:43 -0800, mhoys wrote:

Well, for the normal file operations, especially with small files, Ext4
is quite a bit faster. I've even performed a little benchmark to that
end. Unfortunately, a lot of what I am doing are not "normal file
operations". Virtual machines operate on virtual disks which are
typically 20GB or even more. Databases also have large files which tend
to be accessed using direct I/O. In both cases the ability to defragment
files can lead to palpable performance gains.
I don't know about the "untested" part. XFS was created and put into the
open source by SGI, quite a bit before Ext4 was even conceived. XFS is
definitely older and more mature of the two. Here is a little benchmark
that shows the difference. My /home file system is XFS while the / file
system is Ext4. Here is a little something that illustrates the relative
speeds:

/home partition (XFS):
_____________________


[mgogala@medo Downloads]$ time dd if=oel-6.1.iso of=/dev/null
5902464+0 records in
5902464+0 records out
3022061568 bytes (3.0 GB) copied, 32.1617 s, 94.0 MB/s

real 0m32.206s
user 0m2.869s
sys 0m19.141s
[mgogala@medo Downloads]$ time dd if=oel-6.1.iso of=/dev/null bs=64k
46113+0 records in
46113+0 records out
3022061568 bytes (3.0 GB) copied, 31.8788 s, 94.8 MB/s

real 0m31.949s
user 0m0.048s
sys 0m8.055s
[mgogala@medo Downloads]$ time dd if=oel-6.1.iso of=/dev/null bs=128k
23056+1 records in
23056+1 records out
3022061568 bytes (3.0 GB) copied, 34.0289 s, 88.8 MB/s

real 0m34.483s
user 0m0.051s
sys 0m8.358s
[mgogala@medo Downloads]$ time dd if=oel-6.1.iso of=/dev/null bs=1024k
2882+1 records in
2882+1 records out
3022061568 bytes (3.0 GB) copied, 32.2709 s, 93.6 MB/s

real 0m32.337s
user 0m0.026s
sys 0m9.393s
[mgogala@medo Downloads]$


/tmp (Ext4):
____________


[mgogala@medo tmp]$ time dd if=oel-6.1.iso of=/dev/null
5902464+0 records in
5902464+0 records out
3022061568 bytes (3.0 GB) copied, 28.6371 s, 106 MB/s

real 0m28.675s
user 0m2.270s
sys 0m14.997s
[mgogala@medo tmp]$ time dd if=oel-6.1.iso of=/dev/null bs=64k
46113+0 records in
46113+0 records out
3022061568 bytes (3.0 GB) copied, 29.0088 s, 104 MB/s

real 0m29.067s
user 0m0.084s
sys 0m10.453s
[mgogala@medo tmp]$ time dd if=oel-6.1.iso of=/dev/null bs=128k
23056+1 records in
23056+1 records out
3022061568 bytes (3.0 GB) copied, 27.0044 s, 112 MB/s

real 0m27.065s
user 0m0.054s
sys 0m10.663s
[mgogala@medo tmp]$ time dd if=oel-6.1.iso of=/dev/null bs=1024k
2882+1 records in
2882+1 records out
3022061568 bytes (3.0 GB) copied, 26.5206 s, 114 MB/s

real 0m26.580s
user 0m0.017s
sys 0m11.921s
[mgogala@medo tmp]$


Here is the reason why I selected XFS despite that:

[mgogala@medo tmp]$ filefrag /tmp/oel-6.1.iso
/tmp/oel-6.1.iso: 28 extents found
[mgogala@medo tmp]$ filefrag ~/Downloads/oel-6.1.iso
/home/mgogala/Downloads/oel-6.1.iso: 1 extent found



--
http://mgogala.byethost5.com