<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:georss="http://www.georss.org/georss" xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#" xmlns:media="http://search.yahoo.com/mrss/"
		>
<channel>
	<title>Comments on: Memory Diagnostics &#8211; PGA Part 1</title>
	<atom:link href="http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/feed/" rel="self" type="application/rss+xml" />
	<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/</link>
	<description>Oracle internals, debugging and undocumented features</description>
	<lastBuildDate>Thu, 27 Oct 2011 15:05:30 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.com/</generator>
	<item>
		<title>By: Hans-Peter</title>
		<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/#comment-310</link>
		<dc:creator><![CDATA[Hans-Peter]]></dc:creator>
		<pubDate>Mon, 24 Aug 2009 08:47:42 +0000</pubDate>
		<guid isPermaLink="false">http://oraclue.com/?p=800#comment-310</guid>
		<description><![CDATA[Hi Miladin

Unfortunately I have no direct access myself to the database.
I instructed people on how to get a heap dump for a worker process.
We have a least 4 different ORA-04030&#039;s:
(PLS non-lib hp,pdz2M87_Allocate_Permanent), (sort subheap, sortkey),
(kxs-heap-c,kghsseg: qcstxsInit),
(pga heap,buf_kgcstate).
The physical memory and the ulimits are OK (btw it is AIX 64 bit). 
It is a schema import of a relatively small schema.

I used the available heap dump analyzers.

Regards Hans-Peter]]></description>
		<content:encoded><![CDATA[<p>Hi Miladin</p>
<p>Unfortunately I have no direct access myself to the database.<br />
I instructed people on how to get a heap dump for a worker process.<br />
We have a least 4 different ORA-04030&#8242;s:<br />
(PLS non-lib hp,pdz2M87_Allocate_Permanent), (sort subheap, sortkey),<br />
(kxs-heap-c,kghsseg: qcstxsInit),<br />
(pga heap,buf_kgcstate).<br />
The physical memory and the ulimits are OK (btw it is AIX 64 bit).<br />
It is a schema import of a relatively small schema.</p>
<p>I used the available heap dump analyzers.</p>
<p>Regards Hans-Peter</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: oraclue</title>
		<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/#comment-307</link>
		<dc:creator><![CDATA[oraclue]]></dc:creator>
		<pubDate>Fri, 21 Aug 2009 16:42:30 +0000</pubDate>
		<guid isPermaLink="false">http://oraclue.com/?p=800#comment-307</guid>
		<description><![CDATA[Hi Hans,

Are you getting this error only on impdp? on full impdp or on any impdp ( table level etc )? 

Without knowing much details about your environment I will give you general advice for troubleshooting ORA-4030:

ORA-4030 means that PGA allocation fails due to the following limits. 

1. Physical memory, Swap space ,ulimit etc
2. Oracle PGA limits 
3. Bugs,memory leak , application code 

So first check your OS resources.Are you running out of them?

Then run few generla queries to see PGA usage:

select * from v$pgastat;  for total 

and  

select spid, program, trunc(pga_used_mem/1024/1024) USED,  
trunc(pga_alloc_mem/1024/1024) ALLOC , trunc(pga_max_mem/1024/1024) MAX  
from v$process  
order by pga_alloc_mem desc;

Next would be to do dumps:

 SQL&gt; connect sys as sysdba  
SQL&gt; oradebug setospid SPID  oradebug unlimit  
SQL&gt; oradebug dump heapdump 536870917  
SQL&gt; oradebug tracefile_name 

Also  you can use heapdump analyzer.Tanel got one

http://www.tanelpoder.com/files/scripts/heapdump_analyzer

It is relatively easy to read it.He also got few good articles on memory troubleshooting explained into detail.


I am asssuming that you are running version 10g.There are some unpublished bugs related to expdp/impd.I do not have  call stack from failing impdp and rest of information and cannot make any conclusions.

If you cannot quickly resolve issue can you use exp/imp instead or try to use impdp with few objects.That&#039;s why I have asked you are you using full or partial impdp. 
And  
if you have access to metalink  support then pick the phone and talk to manager and ask for ADVANCED TEAM.They should be able to help you.

Please let me know how it goes.If you need further assistance you can send me info to my email

miladin.modrakovic@gmail.com

I will see what I can do.

Thanks,

Miladin]]></description>
		<content:encoded><![CDATA[<p>Hi Hans,</p>
<p>Are you getting this error only on impdp? on full impdp or on any impdp ( table level etc )? </p>
<p>Without knowing much details about your environment I will give you general advice for troubleshooting ORA-4030:</p>
<p>ORA-4030 means that PGA allocation fails due to the following limits. </p>
<p>1. Physical memory, Swap space ,ulimit etc<br />
2. Oracle PGA limits<br />
3. Bugs,memory leak , application code </p>
<p>So first check your OS resources.Are you running out of them?</p>
<p>Then run few generla queries to see PGA usage:</p>
<p>select * from v$pgastat;  for total </p>
<p>and  </p>
<p>select spid, program, trunc(pga_used_mem/1024/1024) USED,<br />
trunc(pga_alloc_mem/1024/1024) ALLOC , trunc(pga_max_mem/1024/1024) MAX<br />
from v$process<br />
order by pga_alloc_mem desc;</p>
<p>Next would be to do dumps:</p>
<p> SQL&gt; connect sys as sysdba<br />
SQL&gt; oradebug setospid SPID  oradebug unlimit<br />
SQL&gt; oradebug dump heapdump 536870917<br />
SQL&gt; oradebug tracefile_name </p>
<p>Also  you can use heapdump analyzer.Tanel got one</p>
<p><a href="http://www.tanelpoder.com/files/scripts/heapdump_analyzer" rel="nofollow">http://www.tanelpoder.com/files/scripts/heapdump_analyzer</a></p>
<p>It is relatively easy to read it.He also got few good articles on memory troubleshooting explained into detail.</p>
<p>I am asssuming that you are running version 10g.There are some unpublished bugs related to expdp/impd.I do not have  call stack from failing impdp and rest of information and cannot make any conclusions.</p>
<p>If you cannot quickly resolve issue can you use exp/imp instead or try to use impdp with few objects.That&#8217;s why I have asked you are you using full or partial impdp.<br />
And<br />
if you have access to metalink  support then pick the phone and talk to manager and ask for ADVANCED TEAM.They should be able to help you.</p>
<p>Please let me know how it goes.If you need further assistance you can send me info to my email</p>
<p><a href="mailto:miladin.modrakovic@gmail.com">miladin.modrakovic@gmail.com</a></p>
<p>I will see what I can do.</p>
<p>Thanks,</p>
<p>Miladin</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Hans-Peter</title>
		<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/#comment-306</link>
		<dc:creator><![CDATA[Hans-Peter]]></dc:creator>
		<pubDate>Fri, 21 Aug 2009 11:49:46 +0000</pubDate>
		<guid isPermaLink="false">http://oraclue.com/?p=800#comment-306</guid>
		<description><![CDATA[Hi Miladin,

I am trying to diagnose a ORA-4030 that occurs during impdp (trying to allocate 64544 (sort subheap,sort key).
The sort subheap is full with 1182 pieces marked permanent most of the of size 64544 (as the one requested).
Do you have any advice about how to go on? Not much help or Oracle Support.

regards Hans-peter]]></description>
		<content:encoded><![CDATA[<p>Hi Miladin,</p>
<p>I am trying to diagnose a ORA-4030 that occurs during impdp (trying to allocate 64544 (sort subheap,sort key).<br />
The sort subheap is full with 1182 pieces marked permanent most of the of size 64544 (as the one requested).<br />
Do you have any advice about how to go on? Not much help or Oracle Support.</p>
<p>regards Hans-peter</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: oraclue</title>
		<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/#comment-226</link>
		<dc:creator><![CDATA[oraclue]]></dc:creator>
		<pubDate>Tue, 23 Jun 2009 22:14:23 +0000</pubDate>
		<guid isPermaLink="false">http://oraclue.com/?p=800#comment-226</guid>
		<description><![CDATA[Polarski,

Thanks for following my posts.

There is nothing new about dumping and analyzing heaps and subheaps.Plenty of good resources online and they date back many years.

This is just another way to do some of the dumps using undocumented and never before published  option of oradebug command.
Also there is use for any of these commands.Sometimes these commands look useless but that is not the case.You just have to find use for it.Oracle uses it itself.That&#039;s why is undocumented.

Regards,

Miladin]]></description>
		<content:encoded><![CDATA[<p>Polarski,</p>
<p>Thanks for following my posts.</p>
<p>There is nothing new about dumping and analyzing heaps and subheaps.Plenty of good resources online and they date back many years.</p>
<p>This is just another way to do some of the dumps using undocumented and never before published  option of oradebug command.<br />
Also there is use for any of these commands.Sometimes these commands look useless but that is not the case.You just have to find use for it.Oracle uses it itself.That&#8217;s why is undocumented.</p>
<p>Regards,</p>
<p>Miladin</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: oraclue</title>
		<link>http://oraclue.com/2009/06/19/memory-diagnostics-pga-part-1/#comment-225</link>
		<dc:creator><![CDATA[oraclue]]></dc:creator>
		<pubDate>Tue, 23 Jun 2009 22:03:23 +0000</pubDate>
		<guid isPermaLink="false">http://oraclue.com/?p=800#comment-225</guid>
		<description><![CDATA[Dion,

Option unit_test_nolog has many different arguments.I am in process of discovering them ( hard way ).I am planning to write about it in comming posts.Back to your question , I have to check.Did not find any option to dump subheap so far.

Thanks,

Miladin]]></description>
		<content:encoded><![CDATA[<p>Dion,</p>
<p>Option unit_test_nolog has many different arguments.I am in process of discovering them ( hard way ).I am planning to write about it in comming posts.Back to your question , I have to check.Did not find any option to dump subheap so far.</p>
<p>Thanks,</p>
<p>Miladin</p>
]]></content:encoded>
	</item>
</channel>
</rss>

