<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.3.2" -->
<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: Multiple Connections Bringing MySQL to a Halt</title>
	<link>http://www.pythian.com/blogs/878/multiple-connections-bringing-mysql-to-a-halt</link>
	<description>News and views from Pythian DBAs</description>
	<pubDate>Sat, 22 Nov 2008 12:33:54 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.3.2</generator>
		<item>
		<title>By: Jay Pipes</title>
		<link>http://www.pythian.com/blogs/878/multiple-connections-bringing-mysql-to-a-halt#comment-172162</link>
		<dc:creator>Jay Pipes</dc:creator>
		<pubDate>Wed, 19 Mar 2008 13:20:15 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/878/multiple-connections-bringing-mysql-to-a-halt#comment-172162</guid>
		<description>Sounds like an issue of either poor threading code or trying to do too much in a single SQL statement.  Without seeing the code, it's tough to say...

Are they using stored procedures?
Are they pulling large (100,000+ rows) data sets in a buffered result and processing it while maintaining the connection to MySQL?
If they are pulling large data sets, are they pulling with FOR UPDATE?
Are they modifying large ranges of primary keys once processing of the data is done?  This can lead to lock escalation in InnoDB from row to multiple pages (not sure if it will escalate all the way to a table-lock...) depending on how many rows will be updated.

More info would help...</description>
		<content:encoded><![CDATA[<p>Sounds like an issue of either poor threading code or trying to do too much in a single SQL statement.  Without seeing the code, it&#8217;s tough to say&#8230;</p>
<p>Are they using stored procedures?<br />
Are they pulling large (100,000+ rows) data sets in a buffered result and processing it while maintaining the connection to MySQL?<br />
If they are pulling large data sets, are they pulling with FOR UPDATE?<br />
Are they modifying large ranges of primary keys once processing of the data is done?  This can lead to lock escalation in InnoDB from row to multiple pages (not sure if it will escalate all the way to a table-lock&#8230;) depending on how many rows will be updated.</p>
<p>More info would help&#8230;</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Arjen Lentz</title>
		<link>http://www.pythian.com/blogs/878/multiple-connections-bringing-mysql-to-a-halt#comment-172141</link>
		<dc:creator>Arjen Lentz</dc:creator>
		<pubDate>Wed, 19 Mar 2008 11:29:46 +0000</pubDate>
		<guid>http://www.pythian.com/blogs/878/multiple-connections-bringing-mysql-to-a-halt#comment-172141</guid>
		<description>Memory maxes out? Then you might get swapping, which will cause the CPU to max out and indeed the whole system to grind to a halt... keep vmstat running to see.
If the difference is between the 1 and the 5 connections, I presume that some MySQL local buffer settings are set extraordinarily high. Check that.</description>
		<content:encoded><![CDATA[<p>Memory maxes out? Then you might get swapping, which will cause the CPU to max out and indeed the whole system to grind to a halt&#8230; keep vmstat running to see.<br />
If the difference is between the 1 and the 5 connections, I presume that some MySQL local buffer settings are set extraordinarily high. Check that.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
