Raw devices problem

Question

Hello, I made a model on Vmware ESX for testing Oracle RAC, so it's two Centos 5.10 machines with shared disks. I'm using raw devices for ocr, voting disks and data, and without ASMLib library.
After installing clusterware and Oracle (10gR2), everything was OK until now, when after reboot of one node, no more shared disks exist anymore. For every raw disk related comand I got an error:
#raw -qa
Cannot open raw device '/dev/rawctl' (No such file or directory)
#service rawdevices restart
Assigning raw devices:
/dev/raw/raw1 --> /dev/sdb1
Cannot open raw device '/dev/rawctl' (No such file or directory)
Didn't find anything related to this in /var/log's. Could be glad if someone could explain how to fix it.

Answer 1

Yes, but keep in mind that the device tree is not only rebuild after a system restart, but also the order of devices can be different, so for instance /dev/sdc1 may not be the same after a system restart if you attach or remove another device or plug in any USB. That's why UDEV should be configured to use the unique device identifier to map storage devices which require persistent device naming.
For instance you could use the following entry to map e.g. current /dev/sdc1 to /dev/vote1
KERNEL=="sd*", BUS=="scsi", PROGRAM=="/sbin/scsi_id", RESULT=="361a98001686f6959614a453133524171", NAME="vote1", OWNER="oracle", GROUP="oinstall", MODE="0640"
Using raw devices is no longer necessary, not for Oracle, which uses the O_DIRECT flag when opening a device and it does not matter if it is a device or partition. But if you really want to map to raw device names, you could use the following:
ACTION=="add", KERNEL=="sd*", PROGRAM=="/sbin/scsi_id", RESULT=="361a98001686f6959614a453133524171", RUN+="/bin/raw /dev/raw/raw1 %N"