MMapDirectory should only be used on 64bit machines with 64bit JVMs. This is
described in the Javadocs.
The additional sizes are caused by reopens and have their reason in specialties
with the Java VM, described in 2.9/3.0 Javadocs of the class, please take a
look. You can also enable a
Uwe Schindler ,thanks for your reply.
First,I did not use reopens method, even when I start the my program, I can see
this two big anon memory at once.
Second,if I miss unmap,I will see some other index file map to virtual memory
other than anon memory.
-邮件原件-
发件人: Uwe Schindler
Ps
I use Sun JVMs, I will try setUseUnmap(true) to see whether anon memory
disappear.
-邮件原件-
发件人: Uwe Schindler [mailto:u...@thetaphi.de]
发送时间: 2010年1月30日 16:47
收件人: java-user@lucene.apache.org
主题: RE: MMapDirectory need twice more virtual memory than actually need?
MMapDirectory
i try with dbsight and when a try to type retype my passwort:
Can not read Server Configuration file(WEB-INF/data/xsearch-config.xml).
You may avoid this by manually extract the war file.
I already extract the war file ...
Thank u Chris
chrislusf wrote:
For blob, it is not so simple
I found this in http://osdir.com/ml/linux-kernel/2009-10/msg03503.html
If application does mlockall(MCL_FUTURE) it is no longer possible to mmap
file bigger than main memory or allocate big area of anonymous memory
in a thread safe manner. Sometimes it is desirable to lock everything
I am sorry, we know nothing about how the JVM internally handles this. But we
do not force it to map into memory! MMapDirectory just maps but does not call
MappedByteBuffer.load(). So it is completely handled by the kernel when
something is swapped in or not.
I have a server with 32 GB RAM
Are you asking how to get lucene.apache.org out of
http://lucene.apache.org/ or how to get apache.org out of
lucene.apache.org? The getHost() method of java.net.URL will give you
the former. Or use a regexp. I don't know an easy way to do the
latter, but depending on your requirements you could
Ps.
I use tomcat 6
chrislusf wrote:
For blob, it is not so simple since BLOB could contain different file
types, like HTML, pdf, word, zip file type.
So besides getting results out via resultSet.getBlob() function, you
will need to convert the binary stream into simple text strings.