tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Schultz <>
Subject Re: Profiling tomcat with java profiler
Date Fri, 10 Aug 2012 14:46:51 GMT
Hash: SHA1


On 8/10/12 4:25 AM, Ragini wrote:
> On 08/09/2012 08:43 PM, Christopher Schultz wrote: Ragini,
> On 8/9/12 5:51 AM, Ragini wrote:
>>>> what my profiler profiles:- It gives me xml file containing
>>>> list of methods in a proper sequence in which they were
>>>> executed while running the server.
>>>> *My question:* I do the above 3 steps 2 times in exactly
>>>> same manner. So I have two output files say file1 and file2.
>>>> I expected these files to be exactly same because I start the
>>>> server, access the same application and stop the serever. But
>>>> it seems that both files differ. I don't understand why.
>>>> Could you please tell me why does it differ ? Are not methods
>>>> of tomcat executed in exactly same manner for a same action
>>>> ?
>>>> Please find the files attach which I get two different files 
>>>> (file1.xml) and (file2.xml). I have attached differences also
>>>> which it gives. I find these differences by parsing files.
> Instead of dropping two files on the list, why don't you tell us
> what the differences are (in general). It's fine to provide full
> the full data to us, but don't make us wade-through it just ti get
> our bearings.
> -chris
>> ---------------------------------------------------------------------
To unsubscribe, e-mail:
>> For additional commands, e-mail:
> Hi Chris,
> Actually I could not understand the exact differences and why they 
> exist. But in general it seems that sequence of methods which
> starts from org/apache/catalina/startup/HostConfig$DeployWar;"
> name="run", and org/hsqldb/store/HashIndex;" name="getNextLookup
> differes in both the files. I have uploaded the differences file
> ("abranchdifferences") also. So If you kindly have look on few of
> it, you might get some idea.

The JVM is threaded and therefore nondeterministic when it comes to
the exact order of calls, etc. across the whole JVM. If you are
talking about the ordering of certain calls from a /single thread/,
then those ought to be entirely deterministic and I would not expect
to see any deviation between two runs of the same

If you'd like to call our attention to a specific thread of execution
that seems to behave differently in the two separate traces, I might
consider looking. But I'm not even going to open those files until you
get me that far: I'm simply not going to stare at two complete JVM
call traces generated by a "not widely used" profiler for academic

- -chris
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
Comment: GPGTools -
Comment: Using GnuPG with Mozilla -


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message