I’ve been working on a very interesting project: DRaaS (Disaster Recovery as a service)
Yes, that is it. We are configuring a DR on AWS for a huge on-premises SAP system.
Regarding Oracle, we are relying on two technologies:
Oracle DataGuard
So during a test cutover using CloudEndure, we realised that the on-premises server was using udev rules and iSCSI protocol to access the storage.
On AWS EBS you don’t this option so a config change must be made for disks to be available to ASM.
Oracle ASM kfed to the rescue !
But what is kfed ?
It is an Oracle ASM tool to gather ASM disk header information when ASM diskgroups can not be mounted.
Script to fix the udev rules:
_kfed="$(grep ^+ASM /etc/oratab | cut -d\: -f 2)/bin/kfed"
for disk in `blkid | grep oracleasm | awk -F\: '{print $1}'`;do
device_name=`echo $disk | sed 's,/dev/,,g'`
echo -n "KERNEL==\"$device_name\", OWNER=\"grid\", GROUP=\"asmadmin\", MODE=\"0664\", SYMLINK+=\"oracleasm/disks/"
$_kfed read $disk 2>/dev/null | grep dskname | awk '{print $2 "\""}'
done > /etc/udev/rules.d/99-asm.rules
After fixing udev asm rule, you must reload the rules with:
udevadm control --reload-rules udevadm trigger
And check if Oracle ASM disks permissions are correct on /dev
For more information on ASM tools you can check MOS: ASM tools used by Support : KFOD, KFED, AMDU (Doc ID 1485597.1)
