Example (Assume EMC Storage):
# /etc/init.d/oracleasm querydisk -d /dev/emcpowera1
Device "/dev/emcpowera1" is marked an ASM disk with the label "ARC1"
# /etc/init.d/oracleasm querydisk -d ARC1That show major,minor numbers -> 8, 38
Disk "ARC1" is a valid ASM disk on device [8, 38]
So, check on device multi-path:
# ls -la /dev/emcpowera1That made... confuse ;) because on /dev/emcpowera1 show major,minor numbers -> 120, 1
brw------- 1 root root 120, 1 Jun 11 17:01 /dev/emcpowera1
What's that mean? Because the ASMLib scanned on single path to show.
Really?
# ls -la /dev/sd* | grep 38
brw-rw---- 1 root disk 8, 38 Jun 11 17:01 /dev/sdc1
# powermt display dev=emcpoweraThat shows... single paths(sdc,sde,sdg,sdi) on /dev/emcpowera ;)
Pseudo name=emcpowera
Owner: default=SP A, current=SP A Array failover mode: 1
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
1 qla2xxx sdc SP A0 active alive 0 0
1 qla2xxx sde SP A1 active alive 0 0
1 qla2xxx sdg SP B0 active alive 0 0
1 qla2xxx sdi SP B1 active alive 0 0
How we exclude scan on single paths? that's good question ... read on this site.
Excluding Single Path DisksSo, make it on example:
The system administrator configures ASMLib to ignore the single path disks. In the ASMLib configuration, he edits the ORACLEASM_SCANEXCLUDE variable to look like so:
ORACLEASM_SCANEXCLUDE="sdb sdc"
Here, the system administrator has been more specific. ASMLib should ignore exactly the disks /dev/sdb and /dev/sdc. It should not ignore other SCSI disks. While scanning, ASMLib will ignore those paths, only seeing the /dev/multipath disk. Once again, Oracle will use the multipath disk.
Edit /etc/sysconfig/oracleasm file.
# vi /etc/sysconfig/oracleasm
ORACLEASM_SCANEXCLUDE=""
change to =>
ORACLEASM_SCANEXCLUDE="sd"
And then restart:
#/etc/init.d/oracleasm restart
So, Check again:
# /etc/init.d/oracleasm querydisk /dev/emcpowera1
Device "/dev/emcpowera1" is marked an ASM disk with the label "ARC1"
# /etc/init.d/oracleasm querydisk -d ARC1
Disk "ARC1" is a valid ASM disk on device [120, 1]
# ls -al /dev/emcpowera1On ASMLib show major,minor numbers -> 120, 1; that like on multi-path device (asmlib no scan on single paths) ;)
brw------- 1 root root 120, 1 Jun 12 18:33 /dev/emcpowera1
1 comment:
This helped me track down a problem where I could only mount a diskgroup on one node. The problem was that the node on which I had created the ASM disks was seeing them under the major/minor numbers of the /dev/mapper device. On the other nodes, when they scanned for ASM disks, they were seeing them as one of the single paths. I changed this behavior in /etc/sysconfig/oracleasm and it works great!
Post a Comment