This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 57445 - [PERF] Tomcat slow in run mode
Summary: [PERF] Tomcat slow in run mode
Status: RESOLVED INVALID
Alias: None
Product: serverplugins
Classification: Unclassified
Component: Tomcat (show other bugs)
Version: 4.x
Hardware: All Windows XP
: P3 blocker (vote)
Assignee: Pavel Buzek
URL:
Keywords: PERFORMANCE
Depends on:
Blocks:
 
Reported: 2005-04-04 18:21 UTC by _ gtzabari
Modified: 2005-04-13 08:46 UTC (History)
2 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description _ gtzabari 2005-04-04 18:21:47 UTC
dev build 200503301800

When I run my web project (that is, not in debug mode) Tomcat's performance is
*horrible*. Comparing the performance of the integrated Tomcat 5.5 server to my
external one (same machine) the internal one is about 10x slower. There is
absolutely no excuse for this overhead when running in production mode. What gives?
Comment 1 Sherold Dev 2005-04-05 10:04:34 UTC
Marking as WONTFIX.

Please note that the bundled Tomcat server is NOT intended for production but
for development. Although we make some changes in the bundled Tomcat server.xml
configuration file, none of these changes should have such significant impact on
Tomcat performance.

The only thing that should have such impact is the HTTP Monitor. You can disable
it in the Runtime window in the bundled Tomcat properties.
Comment 2 _ rkubacki 2005-04-06 13:02:29 UTC
Gili, can you be more specific wrt what is 10 times slower? Start / deployment /
page processing?

Do you use some customized Tomcat installation or stock distribution? How many apps?

What can be different? JDK, JVM flags, assrtions, verification, security,
network config?
Comment 3 _ gtzabari 2005-04-06 18:09:10 UTC
I simply mean that I hit RUN, the servers starts up fine... what is slow is
using the actual server. That is, from the time it is deployed, hitting pages
off it come back 10x slower than off a normal server. My normal server is a
off-the-shelf Tomcat 5.5.7 installation using the default options. Both Tomcat
installations (internal and external to Netbeans) are running off the same JDK.
Comment 4 _ rkubacki 2005-04-07 10:22:51 UTC
It is possible that JSP compilation is slower as we do not ship with 
jakarta-tomcat-5.5.7/common/lib/jasper-compiler-jdt.jar

I am not sure what is the reasoning behind this.
Comment 5 Sherold Dev 2005-04-07 10:40:22 UTC
We are using Sun JDK compiler instead of Eclipse JDT one. Eclipse JDT compiler
is faster, although I doubt it could be 10 times.

I still think that it is the HTTP Monitor what causes the problem. Could you
please try to disable the HTTP Monitor as described above, to see whether this
is the case?

One part of the HTTP Monitor works as a filter on the server side and processes
every in and out-coming request - it could therefor have a noticeable impact on
request response time.
Comment 6 zikmund 2005-04-07 12:32:07 UTC
Reopening + adding incomplete till we know more details about the problem.
Comment 7 Petr Jiricka 2005-04-07 12:53:12 UTC
Not sure what the problem could be. I know that standalone Tomcat uses Eclipse
compiler, however I think the main reason is that that way, it does not require
a JDK, only JRE. Performance is a secondary reason, AFAIK.

Is this issue reproducible consistently on all machines and platforms? How fast
is it on other machines?
Comment 8 _ gtzabari 2005-04-13 05:12:43 UTC
Sorry for taking so long to reply.

As far as I can see, the HTTP monitor is indeed the culprint. Once I disable it,
restart the webserver and try the webapp again it seems very snappy.

So some food for thought:

1) HTTP Monitor performance should be improved
2) It isn't obvious that one has to restart the server once HTTP Monitor is
disabled for it to *really* get disabled. If you just disable it in properties
without a server restart there is still a noticable performance drop. We should
clarify this behavior somehow if possible.

Let me know if you have any further questions :)
Comment 9 zikmund 2005-04-13 08:04:03 UTC
OK Marking as Invalid then. I'll take care of filing 2 & 1 as a separate bugs.
Comment 10 Sherold Dev 2005-04-13 08:46:30 UTC
Thank you for testing it and clearing it up.

The #2 issue is already filed - issue 44200.