tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Schultz <ch...@christopherschultz.net>
Subject Re: Response Time in Jasper Logs
Date Sun, 05 Jul 2009 13:55:46 GMT
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Eric,

On 7/3/2009 10:42 PM, Robinson, Eric wrote:
> I just want to be sure of one thing. Does the value represent the total
> time from the moment tomcat saw the client's request to the time it
> successfully delivered the full response?

You can always check the source code:
http://svn.apache.org/repos/asf/tomcat/tc6.0.x/tags/TOMCAT_6_0_20/java/org/apache/catalina/valves/AccessLogValve.java

> The reason I ask is that we have some users who are complaining that
> certain requests are taking a long time (up to a minute). Yet when we
> check the jasper logs, we see that 97% of all requests are satisfied in
> < 50ms, and there are no requests that take longer than 20 seconds. So
> I'm trying to reconcile what I'm hearing from users with what I'm seeing
> in the logs.  

Technically speaking, the valve reports the amount of time spent
executing everything "underneath" the valve itself. This may include
other valves, your webapp's filters, and finally your webapp's servlet
(or maybe DefaultServlet or whatever). The bytes are not guaranteed to
have arrived at the client by the time the valve computes the elapsed time.

If a webapp really wants to, it can launch a thread and continue to send
data to the response's output stream, but that will likely cause an
error somewhere down the line, as Tomcat will flush and close the output
stream shortly after it regains control.

The long response times experienced by your clients are more likely to
be due to slow or unreliable networks than anything else: you have the
data to prove that it's not your webapp.

If you have the capability, try benchmarking your webapp from outside
your own network. That will allow you to measure the response time that
the client experiences. Tools such as JMeter are good for this type of
sampling.

Hope that helps,
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkpQsOIACgkQ9CaO5/Lv0PDRsQCcC6beKd95KWkGBIwWYsUYwlrQ
0XYAnjN/g83zPCZpVjctio53Mfv/JxBm
=6NKS
-----END PGP SIGNATURE-----

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org


Mime
View raw message