﻿<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.1.2" -->
<rss version="2.0" 
	xmlns:content="http://purl.org/rss/1.0/modules/content/">
<channel>
	<title>评论: 推荐：深入研究ITL阻塞与ITL死锁</title>
	<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html</link>
	<description>dba on unix</description>
	<pubDate>Fri, 21 Nov 2008 00:29:09 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.1.2</generator>

	<item>
		<title>作者: big_bear</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-389</link>
		<author>big_bear</author>
		<pubDate>Tue, 15 May 2007 02:17:30 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-389</guid>
					<description>支持，写的简明扼要。
能够把死锁产生的trace文件再稍加分析，itl死锁和普通的死锁产生的trace有什么不一样，不过我系统ALERT.LOG文件中发现了死锁日志，怎么去分析TRACE中的文件从而准确判断死锁发生的原因呢？
谢谢！</description>
		<content:encoded><![CDATA[<p>支持，写的简明扼要。<br />
能够把死锁产生的trace文件再稍加分析，itl死锁和普通的死锁产生的trace有什么不一样，不过我系统ALERT.LOG文件中发现了死锁日志，怎么去分析TRACE中的文件从而准确判断死锁发生的原因呢？<br />
谢谢！</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: piner</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-390</link>
		<author>piner</author>
		<pubDate>Tue, 15 May 2007 02:29:58 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-390</guid>
					<description>trace文件都是一样的，从trace不好分别这些类型，只能发现死锁的graph，如果非要找差别，估计只能看下面的Rows waited on，不是数据行(no raw)的等待：
&lt;pre style='background:#e6e6e6;margin: 0cm 0cm 0pt'&gt;
    Deadlock graph:
                           ---------Blocker(s)--------  ---------Waiter(s)---------
    Resource Name          process session holds waits  process session holds waits
    TX-0006001e-0000158d        12     146     X             18     153           S
    TX-00090012-0000157f        18     153     X             17     145           S
    TX-000a0010-00002a98        17     145     X             16     149           S
    TX-0008001d-0000158d        16     149     X             12     146           S
    session 146: DID 0001-000C-00000033     session 153: DID 0001-0012-00000012
    session 153: DID 0001-0012-00000012     session 145: DID 0001-0011-00000074
    session 145: DID 0001-0011-00000074     session 149: DID 0001-0010-000000A3
    session 149: DID 0001-0010-000000A3     session 146: DID 0001-000C-00000033
    Rows waited on:
    Session 153: no row
    Session 145: no row
    Session 149: no row
    Session 146: no row&lt;/pre&gt;</description>
		<content:encoded><![CDATA[<p>trace文件都是一样的，从trace不好分别这些类型，只能发现死锁的graph，如果非要找差别，估计只能看下面的Rows waited on，不是数据行(no raw)的等待：</p>
<pre style='background:#e6e6e6;margin: 0cm 0cm 0pt'>
    Deadlock graph:
                           ---------Blocker(s)--------  ---------Waiter(s)---------
    Resource Name          process session holds waits  process session holds waits
    TX-0006001e-0000158d        12     146     X             18     153           S
    TX-00090012-0000157f        18     153     X             17     145           S
    TX-000a0010-00002a98        17     145     X             16     149           S
    TX-0008001d-0000158d        16     149     X             12     146           S
    session 146: DID 0001-000C-00000033     session 153: DID 0001-0012-00000012
    session 153: DID 0001-0012-00000012     session 145: DID 0001-0011-00000074
    session 145: DID 0001-0011-00000074     session 149: DID 0001-0010-000000A3
    session 149: DID 0001-0010-000000A3     session 146: DID 0001-000C-00000033
    Rows waited on:
    Session 153: no row
    Session 145: no row
    Session 149: no row
    Session 146: no row</pre>
]]></content:encoded>
				</item>
	<item>
		<title>作者: 马艺桐</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-391</link>
		<author>马艺桐</author>
		<pubDate>Tue, 15 May 2007 02:36:54 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-391</guid>
					<description>普通死锁是指的是对相同行的update或者delete的死锁，ITL指的是SEGMENT ITL资源争用引起的死锁。从trace文件中应该可以看到详细信息的：是ITL死锁还是普通的死锁。</description>
		<content:encoded><![CDATA[<p>普通死锁是指的是对相同行的update或者delete的死锁，ITL指的是SEGMENT ITL资源争用引起的死锁。从trace文件中应该可以看到详细信息的：是ITL死锁还是普通的死锁。</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: 马艺桐</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-392</link>
		<author>马艺桐</author>
		<pubDate>Tue, 15 May 2007 02:44:18 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-392</guid>
					<description>也可以使用如下查询，查到数据库当前的ITL等待：
select owner,        object_name,        subobject_name,        object_type,        tablespace_name,        value,        statistic_namefrom   v$segment_statistics where  statistic_name = ’ITL waits’and    value &#62; 0order by value;</description>
		<content:encoded><![CDATA[<p>也可以使用如下查询，查到数据库当前的ITL等待：<br />
select owner,        object_name,        subobject_name,        object_type,        tablespace_name,        value,        statistic_namefrom   v$segment_statistics where  statistic_name = ’ITL waits’and    value &gt; 0order by value;</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: piner</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-393</link>
		<author>piner</author>
		<pubDate>Tue, 15 May 2007 02:45:27 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-393</guid>
					<description>再看了一下，下面的信息也还有点用：

ITL死锁
&lt;coolcode linenum="off"&gt;
last wait for 'enq: TX - allocate ITL entry' blocking sess=0x0x3f2ff954 seq=79 wait_time=2929827 seconds since wait started=213
                name&#124;mode=54580004, usn&lt;&lt;16 &#124; slot=8001d, sequence=158d
    Dumping Session Wait History&lt;/coolcode&gt;

普通死锁与位图索引死锁
&lt;coolcode linenum="off"&gt;    
last wait for 'enq: TX - row lock contention' blocking sess=0x0x3f2ff954 seq=592 wait_time=2929819 seconds since wait started=18
                name&#124;mode=54580004, usn&lt;&lt;16 &#124; slot=7000a, sequence=1570
    Dumping Session Wait History&lt;/coolcode&gt;</description>
		<content:encoded><![CDATA[<p>再看了一下，下面的信息也还有点用：</p>
<p>ITL死锁</p>
<div class="hl-surround"><div class="hl-main">last wait for 'enq: TX - allocate ITL entry' blocking sess=0x0x3f2ff954 seq=79 wait_time=2929827 seconds since wait started=213<br />&nbsp;&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;name|mode=54580004, usn&lt;&lt;16 | slot=8001d, sequence=158d<br />&nbsp;&nbsp; &nbsp;Dumping Session Wait History</div></div>
<p>普通死锁与位图索引死锁</p>
<div class="hl-surround"><div class="hl-main">last wait for 'enq: TX - row lock contention' blocking sess=0x0x3f2ff954 seq=592 wait_time=2929819 seconds since wait started=18<br />&nbsp;&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;name|mode=54580004, usn&lt;&lt;16 | slot=7000a, sequence=1570<br />&nbsp;&nbsp; &nbsp;Dumping Session Wait History</div></div>
]]></content:encoded>
				</item>
	<item>
		<title>作者: piner</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-394</link>
		<author>piner</author>
		<pubDate>Tue, 15 May 2007 02:46:54 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-394</guid>
					<description>[quote]也可以使用如下查询，查到数据库当前的ITL等待：select owner, object_name, subobject_name, object_type, tablespace_name, value, statistic_namefrom v$segment_statistics where statistic_name = ’ITL waits’and value &#62; 0order by value;[/quote]

这个查询的时间跨度太大了，是系统开启以来的。</description>
		<content:encoded><![CDATA[<p>[quote]也可以使用如下查询，查到数据库当前的ITL等待：select owner, object_name, subobject_name, object_type, tablespace_name, value, statistic_namefrom v$segment_statistics where statistic_name = ’ITL waits’and value &gt; 0order by value;[/quote]</p>
<p>这个查询的时间跨度太大了，是系统开启以来的。</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: anysql</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-404</link>
		<author>anysql</author>
		<pubDate>Wed, 16 May 2007 01:12:49 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-404</guid>
					<description>一个ITL只占24个字节, 是可以默认就多给几个的.</description>
		<content:encoded><![CDATA[<p>一个ITL只占24个字节, 是可以默认就多给几个的.</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: NinGoo</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-408</link>
		<author>NinGoo</author>
		<pubDate>Wed, 16 May 2007 05:36:27 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-408</guid>
					<description>呵呵，这个解释得比较清楚了，9i的ITL死锁检测确实是不合理的，所以10g做了修正</description>
		<content:encoded><![CDATA[<p>呵呵，这个解释得比较清楚了，9i的ITL死锁检测确实是不合理的，所以10g做了修正</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: wxy0327</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-447</link>
		<author>wxy0327</author>
		<pubDate>Tue, 22 May 2007 06:51:39 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-447</guid>
					<description>顶</description>
		<content:encoded><![CDATA[<p>顶</p>
]]></content:encoded>
				</item>
	<item>
		<title>作者: big_bear</title>
		<link>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-734</link>
		<author>big_bear</author>
		<pubDate>Fri, 29 Jun 2007 10:05:08 +0000</pubDate>
		<guid>http://www.ixdba.com/html/y2007/m05/101-oracle-itl-deadlock.html#comment-734</guid>
					<description>多谢版主的文章，刚在这里学习完就碰到了这样的情况。附上发生死锁的部分TRACE文件中的信息，从这个文件中也可以看到ITL竞争是很激烈，然后并发大的时候最后导致死锁的。
last wait for 'enq: TX - allocate ITL entry' blocking sess=0x0x10da2a450 seq=34603 wait_time=2930318 secon
ds since wait started=9
                name&#124;mode=54580004, usn</description>
		<content:encoded><![CDATA[<p>多谢版主的文章，刚在这里学习完就碰到了这样的情况。附上发生死锁的部分TRACE文件中的信息，从这个文件中也可以看到ITL竞争是很激烈，然后并发大的时候最后导致死锁的。<br />
last wait for &#8216;enq: TX - allocate ITL entry&#8217; blocking sess=0&#215;0x10da2a450 seq=34603 wait_time=2930318 secon<br />
ds since wait started=9<br />
                name|mode=54580004, usn</p>
]]></content:encoded>
				</item>
</channel>
</rss>
