Re: Tomcat 5.5.25 freezes
We disabled the keep alive connections and our problem is gone... Thanks guys Tony 2008/7/4 Johnny Kewl [EMAIL PROTECTED]: Tony... just another thought... I want to change the philosophy a little If YOU open a stream... ie File f = new File Buffered streams etc... Then make absolutey sure you close them... BUT If you GET a stream from TC... ie OutputStream out = response.getOutputStream(); DONT CLOSE IT... Its an issue we've had, but I didnt see it as a bug, just a feature ;) ie let TC decide when it wants to close its own streams. The theory I have on this is that if you bang the stream shut before all the data is sent... you get the browser waiting for TC problem... and it only shows itself under load... it seems. So its a little contradiction... close only YOUR streams... in theory this is wrong I think, but TC checks it own buffers and will close them at some stage... naturally the stream must not be in a global scope, or TC wont be able to release it it doesnt feel good, but its worth a try... Ok... now I'm fresh out of guesses ;) --- HARBOR : http://www.kewlstuff.co.za/index.htm The most powerful application server on earth. The only real POJO Application Server. See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm --- - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Thanks for your replies, I'll try to disable the keep-alive
connections ... Will get back to you...
One detail I did not mention : I am on a 2003 server ( which does not
make the thread dump that easy ! )
Tony
2008/7/4 Johnny Kewl [EMAIL PROTECTED]:
- Original Message - From: Clovis Wichoski
[EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Friday, July 04, 2008 3:37 AM
Subject: Re: Tomcat 5.5.25 freezes
hi Christophe,
i'm fighting with this problem for months, but i will reply on your thread
for us to make a track of each problem, ok
On Thu, Jul 3, 2008 at 5:49 AM, Christophe Fondacci
[EMAIL PROTECTED] wrote:
Hello,
Might it be related to the problem I described in the thread Tomcat
bottleneck on InternalInputBuffer.parseRequestLine ?
Christophe.
- Original Message - From: Clovis Wichoski
[EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Thursday, July 03, 2008 3:26 AM
Subject: Re: Tomcat 5.5.25 freezes
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED]
wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the
database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Tony, I've never experienced this problem... but heres a guess, and
hopefully a guru can confirm the theory.
When you say 30 minutes, if I'm not mistaken, that is the default time it
takes for a cookie/session to time out... I think. So its somehow related to
that.
I think this is what happens... TC sends up the page and there is a keep
alive header in there... the browser then sends a request and keeps the
connection open TC sends a content size header of say 500 bytes, but
only reads 499... and then its buffer is empty... the browser waits and
waits for that last byte TC never sends it... eventually the cookie
times out, and this prompts TC to drop the connection on the waiting
browser... the browser now displays the data that it got... it looks fine
because the last char was probably a line feed. How can it happen? Thats the
million dollar question... but it maybe something like the content size is
set on the physical file size and a string is sent, that does have the cr.
If the theory is right then you can test it... send some data but set the
content lenght much higher... if it waits... we guessing good ;)
Another test... set the headers to http/1.0... the old standard drops
connections... if it never gets stuck... its related to the http/1.1 spec
which used content size.
It maybe because something else is locking up the file that TC is trying to
read...ie it just cant get the data to send it... or it maybe a bug... or
maybe I should stop guessing ;)
It maybe something tricky... like you updating a file from the EJB... a file
that TC depends on... its locked, or the record is locked... TC cant send...
the EJB is waiting on something that TC must tell it... stuck... and the
browser waits... I think its something like this...
Good Luck...
---
HARBOR : http://www.kewlstuff.co.za/index.htm
The most powerful application server on earth.
The only real POJO Application Server.
See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm
---
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL
Re: Tomcat 5.5.25 freezes
- Original Message - From: Anthony Chamas [EMAIL PROTECTED] To: Tomcat Users List users@tomcat.apache.org Sent: Friday, July 04, 2008 11:01 AM Subject: Re: Tomcat 5.5.25 freezes Thanks for your replies, I'll try to disable the keep-alive connections ... Will get back to you... One detail I did not mention : I am on a 2003 server ( which does not make the thread dump that easy ! ) Its very easy see this link http://www.adventnet.com/products/webnms/help/developer_guide/troubleshooting_tips/creating_threaddump.html Start TC from the command line, the start.BAT... it is destructive, it will take TC down. I dont think disabling Keep alives will do it... I think making the pages http/1.0 will, but the problem is still there, not all the data is getting out. I would start by looking at all the file IO and making absolutely sure the streams are being closed. BTW in the Dos box you can copy the dump... use mark and just press enter, then paste to email. Windows really doesnt like file locks... for example when tomcat is running, try open a log file... you'll see windows complain on notepad. If its only happening on your servlets... I suspect a data size issue. If its happening on static files, like a little old html file, or an image file on disk... something you dont mess with at all, then I think there is a system issue. I would also try turning off any antivirus software on the server... and on the 2000's virus software actually breaks the sockets... so you have to re-apply the latest service pack again... there are articles out there on the subject... Think you going to go, ah silly mistake, or @[EMAIL PROTECTED] ;) Also try different TC versions if you suspect TC... Have fun... now I'm dying to know what the issue is ;) --- HARBOR : http://www.kewlstuff.co.za/index.htm The most powerful application server on earth. The only real POJO Application Server. See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm --- - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Tony, just had another thought When you whack it with Jmeter... open the task manager, and dont watch the memory, with caching and swapping it doesnt mean much... just watch the handles... they should go up and come back to the original levels quickly if they climbing and climbing, and dont recover quickly... something is not closing in the app. Its a rough quicky check without having to open up profilers... ;) Normally those handles will tell you straight away if theres a coding issue in the actual app... they climb and stick up there... --- HARBOR : http://www.kewlstuff.co.za/index.htm The most powerful application server on earth. The only real POJO Application Server. See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm --- - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Tony... just another thought... I want to change the philosophy a little If YOU open a stream... ie File f = new File Buffered streams etc... Then make absolutey sure you close them... BUT If you GET a stream from TC... ie OutputStream out = response.getOutputStream(); DONT CLOSE IT... Its an issue we've had, but I didnt see it as a bug, just a feature ;) ie let TC decide when it wants to close its own streams. The theory I have on this is that if you bang the stream shut before all the data is sent... you get the browser waiting for TC problem... and it only shows itself under load... it seems. So its a little contradiction... close only YOUR streams... in theory this is wrong I think, but TC checks it own buffers and will close them at some stage... naturally the stream must not be in a global scope, or TC wont be able to release it it doesnt feel good, but its worth a try... Ok... now I'm fresh out of guesses ;) --- HARBOR : http://www.kewlstuff.co.za/index.htm The most powerful application server on earth. The only real POJO Application Server. See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm --- - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Thanks for your replies : I'll post the thread dump when I have it..
Cheers,
Anthony
2008/7/3 Clovis Wichoski [EMAIL PROTECTED]:
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED] wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Anthony
*
Here is the architecture :
-struts-, spring-, hibernate- based application with loads of pojos
and db tables (700)
oracle 10g + Tomcat 5.5.25 creating the following DS :
Resource name=jdbc/ORACLEDS scope=Shareable
type=com.mchange.v2.c3p0.ComboPooledDataSource /
ResourceParams name=jdbc/ORACLEDS
parameter namefactory/name
valueorg.apache.naming.factory.BeanFactory/value /parameter
parameter namedriverClass/name
valueoracle.jdbc.driver.OracleDriver/value /parameter
parameter namejdbcUrl/name
valuejdbc:oracle:thin:@10.10.10.10:1521:SID/value /parameter
parameter nameuser/name valueuser/value /parameter
parameter namepassword/name valuepwd/value /parameter
parameter nameinitialPoolSize/name value10/value /parameter
parameter nameminPoolSize/name value10/value /parameter
parameter namemaxPoolSize/name value400/value /parameter
parameter namemaxIdleTime/name value1/value /parameter
parameter nameautoCommitOnClose/name valuefalse/value
/parameter
parameter namenumHelperThreads/name value50/value /parameter
parameter namemaxStatements/name value0/value /parameter
/ResourceParams
The server.xml dealing with the connector :
Connector port=8081 maxHttpHeaderSize=8192
maxThreads=150 minSpareThreads=25 maxSpareThreads=75
enableLookups=false redirectPort=8443 acceptCount=100
connectionTimeout=2 disableUploadTimeout=true /
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Hello,
Might it be related to the problem I described in the thread Tomcat
bottleneck on InternalInputBuffer.parseRequestLine ?
Christophe.
- Original Message -
From: Clovis Wichoski [EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Thursday, July 03, 2008 3:26 AM
Subject: Re: Tomcat 5.5.25 freezes
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED] wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Anthony
*
Here is the architecture :
-struts-, spring-, hibernate- based application with loads of pojos
and db tables (700)
oracle 10g + Tomcat 5.5.25 creating the following DS :
Resource name=jdbc/ORACLEDS scope=Shareable
type=com.mchange.v2.c3p0.ComboPooledDataSource /
ResourceParams name=jdbc/ORACLEDS
parameter namefactory/name
valueorg.apache.naming.factory.BeanFactory/value /parameter
parameter namedriverClass/name
valueoracle.jdbc.driver.OracleDriver/value /parameter
parameter namejdbcUrl/name
valuejdbc:oracle:thin:@10.10.10.10:1521:SID/value /parameter
parameter nameuser/name valueuser/value /parameter
parameter namepassword/name valuepwd/value /parameter
parameter nameinitialPoolSize/name value10/value /parameter
parameter nameminPoolSize/name value10/value /parameter
parameter namemaxPoolSize/name value400/value /parameter
parameter namemaxIdleTime/name value1/value /parameter
parameter nameautoCommitOnClose/name valuefalse/value
/parameter
parameter namenumHelperThreads/name value50/value /parameter
parameter namemaxStatements/name value0/value /parameter
/ResourceParams
The server.xml dealing with the connector :
Connector port=8081 maxHttpHeaderSize=8192
maxThreads=150 minSpareThreads=25 maxSpareThreads=75
enableLookups=false redirectPort=8443 acceptCount=100
connectionTimeout=2 disableUploadTimeout=true /
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
hi Christophe,
i'm fighting with this problem for months, but i will reply on your thread
for us to make a track of each problem, ok
On Thu, Jul 3, 2008 at 5:49 AM, Christophe Fondacci
[EMAIL PROTECTED] wrote:
Hello,
Might it be related to the problem I described in the thread Tomcat
bottleneck on InternalInputBuffer.parseRequestLine ?
Christophe.
- Original Message - From: Clovis Wichoski
[EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Thursday, July 03, 2008 3:26 AM
Subject: Re: Tomcat 5.5.25 freezes
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED]
wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Anthony
*
Here is the architecture :
-struts-, spring-, hibernate- based application with loads of pojos
and db tables (700)
oracle 10g + Tomcat 5.5.25 creating the following DS :
Resource name=jdbc/ORACLEDS scope=Shareable
type=com.mchange.v2.c3p0.ComboPooledDataSource /
ResourceParams name=jdbc/ORACLEDS
parameter namefactory/name
valueorg.apache.naming.factory.BeanFactory/value /parameter
parameter namedriverClass/name
valueoracle.jdbc.driver.OracleDriver/value /parameter
parameter namejdbcUrl/name
valuejdbc:oracle:thin:@10.10.10.10:1521:SID/value /parameter
parameter nameuser/name valueuser/value /parameter
parameter namepassword/name valuepwd/value /parameter
parameter nameinitialPoolSize/name value10/value /parameter
parameter nameminPoolSize/name value10/value /parameter
parameter namemaxPoolSize/name value400/value /parameter
parameter namemaxIdleTime/name value1/value /parameter
parameter nameautoCommitOnClose/name valuefalse/value
/parameter
parameter namenumHelperThreads/name value50/value /parameter
parameter namemaxStatements/name value0/value /parameter
/ResourceParams
The server.xml dealing with the connector :
Connector port=8081 maxHttpHeaderSize=8192
maxThreads=150 minSpareThreads=25 maxSpareThreads=75
enableLookups=false redirectPort=8443 acceptCount=100
connectionTimeout=2 disableUploadTimeout=true /
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
- Original Message -
From: Clovis Wichoski [EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Friday, July 04, 2008 3:37 AM
Subject: Re: Tomcat 5.5.25 freezes
hi Christophe,
i'm fighting with this problem for months, but i will reply on your thread
for us to make a track of each problem, ok
On Thu, Jul 3, 2008 at 5:49 AM, Christophe Fondacci
[EMAIL PROTECTED] wrote:
Hello,
Might it be related to the problem I described in the thread Tomcat
bottleneck on InternalInputBuffer.parseRequestLine ?
Christophe.
- Original Message - From: Clovis Wichoski
[EMAIL PROTECTED]
To: Tomcat Users List users@tomcat.apache.org
Sent: Thursday, July 03, 2008 3:26 AM
Subject: Re: Tomcat 5.5.25 freezes
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED]
wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the
database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Tony, I've never experienced this problem... but heres a guess, and
hopefully a guru can confirm the theory.
When you say 30 minutes, if I'm not mistaken, that is the default time it
takes for a cookie/session to time out... I think. So its somehow related to
that.
I think this is what happens... TC sends up the page and there is a keep
alive header in there... the browser then sends a request and keeps the
connection open TC sends a content size header of say 500 bytes, but
only reads 499... and then its buffer is empty... the browser waits and
waits for that last byte TC never sends it... eventually the cookie
times out, and this prompts TC to drop the connection on the waiting
browser... the browser now displays the data that it got... it looks fine
because the last char was probably a line feed. How can it happen? Thats the
million dollar question... but it maybe something like the content size is
set on the physical file size and a string is sent, that does have the cr.
If the theory is right then you can test it... send some data but set the
content lenght much higher... if it waits... we guessing good ;)
Another test... set the headers to http/1.0... the old standard drops
connections... if it never gets stuck... its related to the http/1.1 spec
which used content size.
It maybe because something else is locking up the file that TC is trying to
read...ie it just cant get the data to send it... or it maybe a bug... or
maybe I should stop guessing ;)
It maybe something tricky... like you updating a file from the EJB... a file
that TC depends on... its locked, or the record is locked... TC cant send...
the EJB is waiting on something that TC must tell it... stuck... and the
browser waits... I think its something like this...
Good Luck...
---
HARBOR : http://www.kewlstuff.co.za/index.htm
The most powerful application server on earth.
The only real POJO Application Server.
See it in Action : http://www.kewlstuff.co.za/cd_tut_swf/whatisejb1.htm
---
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5.5.25 freezes
Hi Anthony,
i have same problem sometimes in version 6.0.16, when the thread stuck
occurs again, execute the follow shell script, please adjust for your
scenario.
#!/bin/bash
today=`date +%Y%m%d%H%M%S`
psId=`/opt/java/jdk1.6.0_06/bin/jps | grep Bootstrap | cut -d' ' -f1`
/opt/java/jdk1.6.0_06/bin/jstack -l $psId
/mnt/logs/stack/stack${today}.txt
execute some times, then post results here, for us to take a look,
regards
Clóvis
On Mon, Jun 30, 2008 at 5:00 AM, Anthony Chamas [EMAIL PROTECTED] wrote:
Hello everyone,
I am having quite a weird situation on a production server : tomcat
seems to be stuck somewhere when handling http requests.
The activity being slow (10 users), some requests take exactly 30
minutes to be processed.
I have put aside the following :
-it happens on any page (not linked to a particular path)
-client issue : sniffing the network did not show anything special :
POST is ok, ACK is received by the client which waits for the server's
reply
-database lock : when that happens, I do not see any lock in the database.
-filters : we tried to deactivate all of them : same result
-I see no errors, warning or something significant in the log files.
I am wondering why I am having the 30 minutes delay... maybe someone
has encountered the same situation... Thanks in advance for helping me
on this one !
Anthony
*
Here is the architecture :
-struts-, spring-, hibernate- based application with loads of pojos
and db tables (700)
oracle 10g + Tomcat 5.5.25 creating the following DS :
Resource name=jdbc/ORACLEDS scope=Shareable
type=com.mchange.v2.c3p0.ComboPooledDataSource /
ResourceParams name=jdbc/ORACLEDS
parameter namefactory/name
valueorg.apache.naming.factory.BeanFactory/value /parameter
parameter namedriverClass/name
valueoracle.jdbc.driver.OracleDriver/value /parameter
parameter namejdbcUrl/name
valuejdbc:oracle:thin:@10.10.10.10:1521:SID/value /parameter
parameter nameuser/name valueuser/value /parameter
parameter namepassword/name valuepwd/value /parameter
parameter nameinitialPoolSize/name value10/value /parameter
parameter nameminPoolSize/name value10/value /parameter
parameter namemaxPoolSize/name value400/value /parameter
parameter namemaxIdleTime/name value1/value /parameter
parameter nameautoCommitOnClose/name valuefalse/value
/parameter
parameter namenumHelperThreads/name value50/value /parameter
parameter namemaxStatements/name value0/value /parameter
/ResourceParams
The server.xml dealing with the connector :
Connector port=8081 maxHttpHeaderSize=8192
maxThreads=150 minSpareThreads=25 maxSpareThreads=75
enableLookups=false redirectPort=8443 acceptCount=100
connectionTimeout=2 disableUploadTimeout=true /
-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
RE: Tomcat 5.5.25 freezes
From: Anthony Chamas [mailto:[EMAIL PROTECTED] I am having quite a weird situation on a production server : tomcat seems to be stuck somewhere when handling http requests. The activity being slow (10 users), some requests take exactly 30 minutes to be processed. I have put aside the following : -it happens on any page (not linked to a particular path) Does it happen on plain content pages as well as dynamic pages (for example, does a CSS ever wait 30 minutes to get served)? Only on dynamic pages, whether or not they touch the database? Only on dynamic pages that touch the database? -database lock : when that happens, I do not see any lock in the database. -filters : we tried to deactivate all of them : same result -I see no errors, warning or something significant in the log files. Next debug step: When it occurs, get a thread dump of your production server. The way to do this varies by OS (which you didn't tell us). It's a non-destructive operation; it will not cause your production server to halt, but it will give you a trace of where each thread is in the system. You might want to do that twice, a few minutes apart, if you reckon something's waiting 30 minutes. That'll give you a lot of data to look over. I'd be hunting for threads that were waiting on locks or waiting in the database connector. Expect lots of waiting threads - most of them will be idle, waiting to process a request. This is not an error condition, it's normal! If you don't get any joy looking at the thread dump (and they're not the easiest things in the world to interpret), some of the folks on-list may be willing to help. - Peter - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: Tomcat 5.5.25 freezes
Also make sure GC is not an issue. If the occurrence is random and there is no lock in threads / database , then GC is usually the cause. You can enable JMX and monitor it using Jconsole. Cheers Suren -Original Message- From: Peter Crowther [mailto:[EMAIL PROTECTED] Sent: Monday, June 30, 2008 1:53 PM To: 'Tomcat Users List' Subject: RE: Tomcat 5.5.25 freezes From: Anthony Chamas [mailto:[EMAIL PROTECTED] I am having quite a weird situation on a production server : tomcat seems to be stuck somewhere when handling http requests. The activity being slow (10 users), some requests take exactly 30 minutes to be processed. I have put aside the following : -it happens on any page (not linked to a particular path) Does it happen on plain content pages as well as dynamic pages (for example, does a CSS ever wait 30 minutes to get served)? Only on dynamic pages, whether or not they touch the database? Only on dynamic pages that touch the database? -database lock : when that happens, I do not see any lock in the database. -filters : we tried to deactivate all of them : same result -I see no errors, warning or something significant in the log files. Next debug step: When it occurs, get a thread dump of your production server. The way to do this varies by OS (which you didn't tell us). It's a non-destructive operation; it will not cause your production server to halt, but it will give you a trace of where each thread is in the system. You might want to do that twice, a few minutes apart, if you reckon something's waiting 30 minutes. That'll give you a lot of data to look over. I'd be hunting for threads that were waiting on locks or waiting in the database connector. Expect lots of waiting threads - most of them will be idle, waiting to process a request. This is not an error condition, it's normal! If you don't get any joy looking at the thread dump (and they're not the easiest things in the world to interpret), some of the folks on-list may be willing to help. - Peter - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
