<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.6.5" -->
<rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	>
<channel>
	<title>Comments on: Running RAC and ASM on Linux (Finally!)</title>
	<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux</link>
	<description>News and views from Pythian DBAs</description>
	<pubDate>Thu,  4 Dec 2008 21:51:29 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.6.5</generator>
		<item>
		<title>By: sdquirra</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-256908</link>
		<dc:creator>sdquirra</dc:creator>
		<pubDate>Wed, 13 Aug 2008 09:29:40 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-256908</guid>
		<description>hi,
using /dev/sd* devices for ocr and voting disk, i'm getting a "the location /dev/sdb1, entered for the oracle cluster registry (ocr) is not shared across all the nodes in the cluster....".
This post also http://forums.oracle.com/forums/thread.jspa?threadID=641837&#38;tstart=0&#38;messageID=2474588 says that, for ocr and voting disk, 10g keep using raw dev or cluster fs.
It is correct?
thnx</description>
		<content:encoded><![CDATA[<p>hi,<br />
using /dev/sd* devices for ocr and voting disk, i&#8217;m getting a &#8220;the location /dev/sdb1, entered for the oracle cluster registry (ocr) is not shared across all the nodes in the cluster&#8230;.&#8221;.<br />
This post also <a href="http://forums.oracle.com/forums/thread.jspa?threadID=641837&amp;tstart=0&amp;messageID=2474588" rel="nofollow">http://forums.oracle.com/forums/thread.jspa?threadID=641837&amp;tstart=0&amp;messageID=2474588</a> says that, for ocr and voting disk, 10g keep using raw dev or cluster fs.<br />
It is correct?<br />
thnx</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Grégory</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-170481</link>
		<dc:creator>Grégory</dc:creator>
		<pubDate>Thu, 13 Mar 2008 12:50:49 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-170481</guid>
		<description>the answer with Linux is :
a- prefer using ASMLib. In particular is enables to discover the disks on all nodes without partprobe and it is optimized for ASM.
b- If you cannot, use the block device, i.e /dev/sdb* and NOT raw devices</description>
		<content:encoded><![CDATA[<p>the answer with Linux is :<br />
a- prefer using ASMLib. In particular is enables to discover the disks on all nodes without partprobe and it is optimized for ASM.<br />
b- If you cannot, use the block device, i.e /dev/sdb* and NOT raw devices</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Cristian Cudizio</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-170337</link>
		<dc:creator>Cristian Cudizio</dc:creator>
		<pubDate>Wed, 12 Mar 2008 20:28:20 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-170337</guid>
		<description>sorry, i don't understand one thing: can i use directly /dev/sdb* for asm disks with 10gR2? it seems to me that also in 11g documentation Oracle says to use ASMLib but i can't find ASMLib very useful. Morehover i've met problem described in Metalink Note 457369.1 that suggest to use asm_diskstring = /dev/oracleasm/disks/* but devices files under /dev/oracleasm/disks/* have same major e minor numbers of original block devices (/dev/sdb*) so i think that is the same think. So my question is: does oracle support the use of asm_diskstring=/dev/sdb* ?

Compliments for the post. 

Thanks,
 Cristian</description>
		<content:encoded><![CDATA[<p>sorry, i don&#8217;t understand one thing: can i use directly /dev/sdb* for asm disks with 10gR2? it seems to me that also in 11g documentation Oracle says to use ASMLib but i can&#8217;t find ASMLib very useful. Morehover i&#8217;ve met problem described in Metalink Note 457369.1 that suggest to use asm_diskstring = /dev/oracleasm/disks/* but devices files under /dev/oracleasm/disks/* have same major e minor numbers of original block devices (/dev/sdb*) so i think that is the same think. So my question is: does oracle support the use of asm_diskstring=/dev/sdb* ?</p>
<p>Compliments for the post. </p>
<p>Thanks,<br />
 Cristian</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Oracle Clusterware on RHEL5/OEL5 with udev and multipath : Ardent Performance Computing</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-161672</link>
		<dc:creator>Oracle Clusterware on RHEL5/OEL5 with udev and multipath : Ardent Performance Computing</dc:creator>
		<pubDate>Thu, 14 Feb 2008 05:38:58 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-161672</guid>
		<description>[...] no permissions.d directory on RHEL5 and OEL5. Last September, Grégory Guillou from Pythian blogged about installing on Red Hat 5 and referenced a post on his French blog that shows how to setup a RULES file for SCSI block [...]</description>
		<content:encoded><![CDATA[<p>[&#8230;] no permissions.d directory on RHEL5 and OEL5. Last September, Grégory Guillou from Pythian blogged about installing on Red Hat 5 and referenced a post on his French blog that shows how to setup a RULES file for SCSI block [&#8230;]</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Grégory</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-105173</link>
		<dc:creator>Grégory</dc:creator>
		<pubDate>Tue, 25 Sep 2007 03:19:21 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-105173</guid>
		<description>Gawie,

Lets start with the bad news : My original post &lt;a href="http://arkzoyd.blogspot.com/2007/09/oracle11g-sur-linux-la-fin-des-raw.html" rel="nofollow"&gt; about raw devices is in French&lt;/a&gt; !

What it says is that with RHEL5/OEL5, the /etc/udev/permissions.d directory doesn't exist anymore and you have to create a file under /etc/udev/rules.d to set the permissions on boot ; you should be able to get the content of the file from the post even if it's in French !

Regarding Oracle recommendation about "raw vs block" devices you'll find a couple of informations in different 11g docs :

1- (About 10g first) &lt;a href="http://download.oracle.com/docs/cd/B28359_01/rac.111/b28252/preparing.htm#CEGGDJHH" rel="nofollow"&gt;This section&lt;/a&gt; explains that starting with clusterware 10.2 and RHEL4, you can use block (instead of raw) devices to store the OCR and the voting disk.

2- (11g) &lt;a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIBCBJ" rel="nofollow"&gt;In this section&lt;/a&gt;, the doc says that Oracle recommends using block rather that raw devices on Linux. BTW, &lt;a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIJJEH" rel="nofollow"&gt;this section&lt;/a&gt; explains how to configure block devices while storing one file per block device (As you use to do with raw devices).

3- (10g and 11g) &lt;a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIFHAB" rel="nofollow"&gt;This section&lt;/a&gt; explains that Oracle recommends using block devices with ASMLib to create ASM disks. &lt;a href="http://www.oracle.com/technology/tech/linux/asmlib/index.html" rel="nofollow"&gt;This OTN page&lt;/a&gt; contains the supported release of ASMLib, including the one for RHEL4 and RHEL5. Well (Shame on me !) this is not what I did ! I've changed the asm_diskstring parameter and I've bound the ASM DISKS to the block devices. I probably shouldn't ! I'm pretty sure the problem I've encountered with lsdsk will be solved by doing the setup this way : I'll check soon !

Anyway 1 and 3 show that you don't need the raw devices anymore (even with 10g and RHEL4). But it doesn't prevent you to set the permission via the "udev" dynamic device naming.</description>
		<content:encoded><![CDATA[<p>Gawie,</p>
<p>Lets start with the bad news : My original post <a href="http://arkzoyd.blogspot.com/2007/09/oracle11g-sur-linux-la-fin-des-raw.html" rel="nofollow"> about raw devices is in French</a> !</p>
<p>What it says is that with RHEL5/OEL5, the /etc/udev/permissions.d directory doesn&#8217;t exist anymore and you have to create a file under /etc/udev/rules.d to set the permissions on boot ; you should be able to get the content of the file from the post even if it&#8217;s in French !</p>
<p>Regarding Oracle recommendation about &#8220;raw vs block&#8221; devices you&#8217;ll find a couple of informations in different 11g docs :</p>
<p>1- (About 10g first) <a href="http://download.oracle.com/docs/cd/B28359_01/rac.111/b28252/preparing.htm#CEGGDJHH" rel="nofollow">This section</a> explains that starting with clusterware 10.2 and RHEL4, you can use block (instead of raw) devices to store the OCR and the voting disk.</p>
<p>2- (11g) <a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIBCBJ" rel="nofollow">In this section</a>, the doc says that Oracle recommends using block rather that raw devices on Linux. BTW, <a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIJJEH" rel="nofollow">this section</a> explains how to configure block devices while storing one file per block device (As you use to do with raw devices).</p>
<p>3- (10g and 11g) <a href="http://download.oracle.com/docs/cd/B28359_01/install.111/b28263/stor_rac.htm#BABIFHAB" rel="nofollow">This section</a> explains that Oracle recommends using block devices with ASMLib to create ASM disks. <a href="http://www.oracle.com/technology/tech/linux/asmlib/index.html" rel="nofollow">This OTN page</a> contains the supported release of ASMLib, including the one for RHEL4 and RHEL5. Well (Shame on me !) this is not what I did ! I&#8217;ve changed the asm_diskstring parameter and I&#8217;ve bound the ASM DISKS to the block devices. I probably shouldn&#8217;t ! I&#8217;m pretty sure the problem I&#8217;ve encountered with lsdsk will be solved by doing the setup this way : I&#8217;ll check soon !</p>
<p>Anyway 1 and 3 show that you don&#8217;t need the raw devices anymore (even with 10g and RHEL4). But it doesn&#8217;t prevent you to set the permission via the &#8220;udev&#8221; dynamic device naming.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Gawie</title>
		<link>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-105047</link>
		<dc:creator>Gawie</dc:creator>
		<pubDate>Mon, 24 Sep 2007 22:07:37 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/613/running-rac-and-asm-on-linux#comment-105047</guid>
		<description>Hi Gregory,

Could you point me to the previous posts regarding etc/udev/rules.d and the fact that no RAW devices are needed anymore.

Thank you for the information in this post
Gawie</description>
		<content:encoded><![CDATA[<p>Hi Gregory,</p>
<p>Could you point me to the previous posts regarding etc/udev/rules.d and the fact that no RAW devices are needed anymore.</p>
<p>Thank you for the information in this post<br />
Gawie</p>
]]></content:encoded>
	</item>
</channel>
</rss>
