RE: fdisk, device mapper and asmlib
Date: Fri, 24 Feb 2012 19:34:11 +0000
You should put in a multipath_bindings file and take precautions that you are mapping the same devices to the same ID's every time. Sometimes you can get into cases where the devices map differently on reboot. -----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Radoulov, Dimitre Sent: Friday, February 17, 2012 4:09 PM
To: D'Hooge Freek
Subject: Re: fdisk, device mapper and asmlib
Thank you Freek,
I'll discuss your suggestions and Li Li's with the storage team.
On 17/02/2012 22:05, D'Hooge Freek wrote:
> You probably need to rescan the partition table from the second node.
> You can do this by using blockdev --rereadpt<single path device>
> Once you have done this, it could be you still need to reread the partition table on the multipath device, which can be done using partprobe. After that it could be necessary to use kpartx to create the new partition.
> I never found why sometimes the multipath and / or the multipath partition steps needed to be done and why the sometimes where directly correct after using the blockdev command on one of the single path devices.
> I can only guess it has something to do with caching.
> Freek D'Hooge
> Oracle Database Administrator
> email: freek.dhooge_at_uptime.be
> tel +32(0)3 451 23 82
> disclaimer: www.uptime.be/disclaimer
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Radoulov, Dimitre
> Sent: vrijdag 17 februari 2012 21:42
> To: ORACLE-L
> Subject: Re: fdisk, device mapper and asmlib
> Just to add that kpartx and partprobe were executed on the second node
> after the partition was recreated.
> On 17/02/2012 21:34, Radoulov, Dimitre wrote:
>> Hi all,
>> again a 2 node RAC 184.108.40.206.1 setup on RHEL 5.7
>> asmlib oracleasm-2.6.18-274.17.1.el5-2.0.5-1.el5
>> (kernel in use oracleasm-2.6.18-274.17.1.el5)
>> Linux native multipath - device mapper
>> asm scan order: "mpath_string dm"
>> asm exclude: "sd"
>> We're facing the following issue: