100% CPU usage after resume WinXP from stand-by: appears to be in MINA
----------------------------------------------------------------------
Key: DIRSERVER-608
URL: http://issues.apache.org/jira/browse/DIRSERVER-608
Project: Directory ApacheDS
Type: Bug
Components: core
Versions: 1.0-RC1
Environment: Windows XP 2002 SP2, JDK 1.4.2_04-b05
Reporter: Norbert Reilly
I've noticed that when ApacheDS is running on my laptop and I close the lid and
later open it that the ApacheDS java process is using 100% CPU. When logging at
the DEBUG level the 100% CPU usage seems to coincide with messages like this
(the 20 second periodic sync):
============================
2006-04-19 12:46:08,833 101464 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN 'ou=system'
2006-04-19 12:46:08,883 101514 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN
'uid=admin,ou=system'
2006-04-19 12:46:08,943 101574 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN
'uid=admin,ou=system'
2006-04-19 12:46:09,013 101644 [SynchWorkerThread]
(org.apache.directory.server.core.authn.AuthenticationService:248) DEBUG
org.apache.directory.server.core.authn.AuthenticationService - Testing if
entry name = 'ou=system' exists
2006-04-19 12:46:09,063 101694 [SynchWorkerThread]
(org.apache.directory.server.core.partition.DefaultDirectoryPartitionNexus:722)
DEBUG org.apache.directory.server.core.partition.DefaultDirectoryPartitionNexus
- Check if DN 'ou=system' exists.
2006-04-19 12:46:09,123 101754 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN
'ou=services,ou=configuration,ou=system'
2006-04-19 12:46:09,223 101854 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN 'ou=system'
2006-04-19 12:46:09,304 101935 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN
'ou=interceptors,ou=configuration,ou=system'
2006-04-19 12:46:09,424 102055 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN 'ou=system'
2006-04-19 12:46:09,504 102135 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN
'ou=configuration,ou=system'
2006-04-19 12:46:09,604 102235 [SynchWorkerThread]
(org.apache.directory.shared.ldap.name.DnParser:184) DEBUG
org.apache.directory.shared.ldap.name.DnParser - Parsing DN 'ou=system'
============================
This is a dump of the threads when I pause the debugger:
[EMAIL PROTECTED], priority=5, in group 'main', status: 'RUNNING'
[EMAIL PROTECTED], priority=5, in group 'main', status: 'WAIT'
wait():-1, Object.java
run():155, Service.java
run():534, Thread.java
[EMAIL PROTECTED], priority=5, in group 'main', status: 'RUNNING'
poll0():-1, WindowsSelectorImpl.java
poll():263, WindowsSelectorImpl.java
access$400():245, WindowsSelectorImpl.java
doSelect():128, WindowsSelectorImpl.java
lockAndDoSelect():59, SelectorImpl.java
select():70, SelectorImpl.java
select():74, SelectorImpl.java
run():326, SocketAcceptorDelegate.java
Signal [EMAIL PROTECTED] daemon, priority=10, in group 'system', status:
'RUNNING'
[EMAIL PROTECTED] daemon, priority=8, in group 'system', status: 'WAIT'
wait():-1, Object.java
remove():111, ReferenceQueue.java
remove():127, ReferenceQueue.java
run():159, Finalizer.java
Reference [EMAIL PROTECTED] daemon, priority=10, in group 'system', status:
'WAIT'
wait():-1, Object.java
wait():429, Object.java
run():115, Reference.java
The busy loop is in
org.apache.mina.transport.socket.nio.support.SocketAcceptorDelegate.Worker.run()
and appears to be because the line:
int nKeys = selector.select();
never blocks after the resume, and hence the run() method becomes a busy loop.
This may well be a bug in the JDK's nio package but a quick web search didn't
turn up a match. At any rate it seems like a problem that will need to be
worked around in MINA until the JDK is patched (if my diagnosis is correct) at
it seems the selector is rendered inoperable by the close-lid/resume sequence.
I presume the same problem could be triggered on a desk-top machine with enough
understanding about what WinXP does during a suspend, and why this upsets the
NIO package.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira
