10g Release 2

This is as much a note to myself how to do this in the future as it is something hopefully worth reading for you. The requirement has been precise as always: migrate a database from 10.2 on SPARC to 11.2 on Linux. In the process, go from Veritas to ASM and make it quick!

I like short briefings but this was too short. Since the database was reasonably large I opted for the transportable tablespace approach, however I now think that a massively parallel impdp with network_link could have saved me quite a bit of time.

The following is by no means  the complete story, but hopefully gives you an idea how to do these things. Always check, and document, then test (rinse and repeat). Only when proper signoff is received should you try such a process in production. Remember to script it and have at least one clean run of the scripts! This process is not super-quick, if you have low downtime requirements then consider Streams or better: Golden Gate for the process.

This post is long overdue, and it’s as much a reminder to myself as it is a post for you to find when scratching your head.

Here’s the situation: Oracle Restart on an Oracle Enterprise Linux 5.5 system with ASMLib in use. I have installed 11.2.0.2 plus the latest PSU to bring database and Grid Infrastructure to 11.2.0.2.5-so far so good. I have used the non-GUI approach and everything seemed ok until I tried to create a database for some benchmarking. Here’s the surprising output:

During some recent work I did involving a stretched 11.2.0.2 RAC for a SAP implementation at a customer site I researched TimeFinder/Clone backups. As part of this exercise I have been able to experiment with RHEL (OEL) 5.6 and the new device mapper multipath package on the mount host. I have been very pleasantly surprise about this new feature which I’d like to share.

Background of this article

Device Mapper Multipath is the “native” Linux multipathing software, as opposed to vendor-supplied multipathing such as EMC’s Power Path or Hitachi’s HDLM.

My customer’s setup is rather unique for a SAP environment as it uses Oracle Enterprise Linux and not Solaris/SPARC or AIX on the Power platform with an active/passive solution. Well if that doesn’t make it sound unique, the fact that there is a plan to run Oracle 11.2.0.2 RAC potentially across sites using ASM and ACFS certainly makes this deployment stand out from the rest.

The reason I mention this is simple-it requires a lot of engineering effort to certify components for this combination. For example: it was not trivial to get vendor support for Solutions Enabler the storage engineering uses for connectivity with the VMAX arrays, and so on. After all, Oracle Enterprise is a fairly new platform and Red Hat certainly has an advantage when it comes to vendor certification.

What hasn’t been achieved was a certification of the EMC Power Path software for use with SAP on Linux, for reasons unknown to me. The result was simple: the setup will use the device-mapper multipath package that comes with the Linux distribution.

Configuring multipath

Now with this said I started looking at MOS to get relevant support notes about Linux’s native multipath and found some. The summary of this research is available on this blog, I have written about it in these articles:

• http://martincarstenbach.wordpress.com/2010/11/16/configuration-device-m...
• http://martincarstenbach.wordpress.com/2011/01/14/adding-storage-dynamic...
• http://martincarstenbach.wordpress.com/2009/12/04/let-there-be-multipath...

What I didn’t know up to date was the fact that the multipath.conf file allows you to define the ownership and mode of a device. As an ASM user this is very important to me. Experience taught me that incorrect permissions are one of the main reasons for ASM failing to start. Remember that root owns block devices by default.

Consider this example:

multipaths {
...
multipath {
wwid        360a98000486e58526c34515944703277
alias       ocr01
mode        660
uid         501
gid         502
}
...
}

The above entry in the multipaths section translates into English as follows:

• If you encounter a device with the WWID 36a…277
• Give it an alias name of “OCR01” in /dev/mapper
• Set the mode to 660 (i.e. rw-rw—)
• Assign device ownership to the user with UID 501 (maps to “grid” on my system)
• Assign the group of the device to 502 (maps to “asmdba” on my system)

The path settings are defined globally and don’t need to be mentioned explicitly for each device unless you prefer to override them. I like to use an alias although it isn’t really necessary since ASM relies on a 4k header in a block device to store its identity. If you don’t chose to alias a device I recommend you use the user friendly name instead, mainly for aesthetic reasons.

Why is this really cool? For me it implies two things:

• I can now be independent of ASMLib which provided device name stability and set the permissions on the block devices correctly for ASM
• I don’t need to create udev rules to set the permissions (or /etc/rc.local or whichever other way you chose before to set permissions)

Nice! One less headache, as I have to say that I didn’t really like udev…