RE: Performance issue

[Bill Sweeney]

All the posts so far are right - you'll need to profile the app ;-)

A few tuning thoughts as you zero in on the app issues: 

Check your logs.  Are you getting connection refused errors at peak when the pages 
take so long to load?  Modify your connector in your server.xml.  You can up the 
maxProcessors=254 before needing to grab the source.  If there is a need to up the 
wait queue you can up acceptCount=300.  Also lessen the cycles for DNS lookups by 
enableLookups=false.

Are you starved for DB connections?  Add more in your [app_context].xml (20 is default 
I think)
   parameter
   namemaxActive/name
   value80/value
 /parameter
 parameter
   namemaxIdle/name
   value80/value
 /parameter


Is the compiler and execution servlet in developer mode?  Is Tomcat looking for 
changes in the JSPs at every access?  Look in web.xml and make

init-param
param-namedevelopment/param-name
param-valuefalse/param-value
/init-param
Up your checkinterval or set 'reloading' to false as well.

You said I have 128M initial memory, and 2GB max memory for JVM.  In production 
consider giving Tomcat all the memory at initial startup that you intend to give it at 
max.  We saw evidence of memory fragmentation that was a performance hit in our 
testing.  But again, you need to profile your app.  Are you swapping?  

You said we're using a wrapper to restart it after it crashes.  Are you precompiling 
the app, or compiling the pages on access?  If not precompiling, you have beefy iron 
but enough restarts and that is a lot of cycles at peak.

Good Luck - 

Bill








-Original Message-
From: QM [mailto:[EMAIL PROTECTED]
Sent: Thursday, October 07, 2004 7:46 AM
To: Tomcat Users List
Subject: Re: Performance issue


On Thu, Oct 07, 2004 at 11:40:50AM +0800, Faisal Abdullah wrote:
: I think its probably its how the app manages connection, and probably GC?
: DBCP and Cayote HTTP1/1 isn't configured by the way. Would these help?
: If I configure DBCP, would it effect the code?

DBCP may affect the overall app performance: it keeps a set of open
database connections, instead of initiating a new connection each time
someone calls the database.  The creation/teardown of a network
connection can be expensive.

Right now it seems you're guessing at the cause.  (You're likely
correct, but still guessing.)  Analyze the app with a profiler to
determine exactly what's going on.  A profiler will show you what method
calls are taking the most time, what's in the heap, whether GC is the
cause of the performance problem, etc.

-QM

-- 

software  -- http://www.brandxdev.net
tech news -- http://www.RoarNetworX.com


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Tomcat and encryption cards

[Bill Sweeney]

Hello all - 

Does anyone know if Tomcat supports the Sun Crypto Accelerator card?  

We think that since Tomcat uses JSSE we should be all set but can't find anyone 
actually doing it.

Bill


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

tomcat kernel32.dll fault

[Bill Sweeney]

Hello - 

We are running a load test on our app and have followed many of the performance and 
memory tweaks and testing advice from the Users List - thanks to all who contribute.

We hit a problem where under load (50 users generated by loadrunner) the Windows 2003 
Server running Tomcat4.1.30 will throw the following error to the windows event log:

: 41 70 70 6c 69 63 61 74   Applicat
0008: 69 6f 6e 20 46 61 69 6c   ion Fail
0010: 75 72 65 20 20 74 6f 6d   ure  tom
0018: 63 61 74 2e 65 78 65 20   cat.exe 
0020: 31 2e 31 2e 30 2e 30 20   1.1.0.0 
0028: 69 6e 20 6b 65 72 6e 65   in kerne
0030: 6c 33 32 2e 64 6c 6c 20   l32.dll 
0038: 35 2e 32 2e 33 37 39 30   5.2.3790
0040: 2e 30 20 61 74 20 6f 66   .0 at of
0048: 66 73 65 74 20 30 30 30   fset 000
0050: 30 31 37 62 31017b1

... and the tomcat exits.  This is reproducable on 2 different Win2003 servers. On any 
given machine the offset is the same every time the error occurs.

Googled the error a few ways and can't find any info.

Has anyone encountered this? or have a suggestion to try and locate the cause?

Thanks - 

Bill


 


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]