You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by Daniel Lopez <gr...@dev.java.net> on 2009/04/05 20:04:19 UTC
Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Hi there,
I have an application that uses Tomcat as embedded engine and so far I
was using version 5.5 of Tomcat. I wanted to upgrade to version 6.0.18
and I've found some issues due to the different behaviour and I don't
see any difference in the API or articles I've found so I wonder what
I'm doing wrong.
The first problem is that, after using Server.start(), unless I have
another living thread, the program will exit even though I have the
container started and supposedly listening on the port. I have tested
that the server indeed works and listens to that port, but when my
other threads finish, the container simply exists without a trace.
I've seen a thread in the tomcat-dev list ("embedded tomcat 6" 14 Feb
2008) that seems to imply this is normal with Tomcat 6, but in Tomcat
5.5 this did not happen. Is this normal?
On the other hand, if I start the server with a GUI window, so the
container remains alive due to the GUI thread, stopping the server
does not really work, as even though it says that the Connector
stopped, I can still access the server for some seconds and then it
starts answering "resource not available", but the thing is, it still
answers to the port. And I starting the server again does not "fix
it". On the other hand, with Tomcat 5.5 I am able to stop the server,
and it stops answering at the port, so I am also able to start it
again. Using exactly the same code.
The code looks like this:
----------------------------------
this.theEmbedded = new Embedded();
this.theEmbedded.setCatalinaHome(new File("tomcat_6").getAbsolutePath());
// Add the engine
this.engine = this.theEmbedded.createEngine();
this.engine.setName("Catalina");
this.engine.setDefaultHost("localhost");
// Create the host
this.host = this.theEmbedded.createHost("localhost", ".");
this.engine.addChild(this.host);
this.theContext = this.theEmbedded.createContext("/",
this.theConfiguration.getBasePath());
this.theContext.setReloadable(false);
this.host.addChild(this.theContext);
this.theEmbedded.addEngine(this.engine);
// Add a conector
this.connector =
this.theEmbedded.createConnector((java.net.InetAddress) null,
this.port, false);
this.theEmbedded.addConnector(this.connector);
this.theEmbedded.start();
----------------------------------
Stopping it is quite simple:
this.theEmbedded.stop();
----------------------------------
So, is the Tomcat 6 embedded engine broken or do I have to change the
code? Nothing in the API seemed to indicate any changes needed but...
who knows? :)
Thanks for your help,
D.
PD: I can provide more information like libraries used and log traces,
but I did not want to make this message too long, in case it was
something obvious I had missed.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
RE: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by "Caldarale, Charles R" <Ch...@unisys.com>.
> From: André Warnier [mailto:aw@ice-sa.com]
> Subject: Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
>
> I am surprised that you would not have heard of stealth technology.
> What do they call a stealthy Tomcat ? a Raptor ?
I think it will be the Lightning II; the US Navy passed on the Raptor (puts on a good show at Oshkosh, though).
- Chuck
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by János Löbb <ja...@yale.edu>.
On Apr 7, 2009, at 7:57 AM, André Warnier wrote:
> János Löbb wrote:
>> On Apr 6, 2009, at 4:10 AM, André Warnier wrote:
>>> Caldarale, Charles R wrote:
>>> [...]
>>>> The metaphysical implications of existing without a trace are
>>>> rather intriguing...
>>> I am surprised that you would not have heard of stealth technology.
>>> What do they call a stealthy Tomcat ? a Raptor ?
>> No.... that must be a Schrödingercat :)
> I am sorry, but I will have to disagree.
> A Schrödinger Tomcat would be one that exists multiple times, in
> quantum superposition. Which would probably create problems with all
> of them trying to listen on the same TCP ports. Unless of course the
> listening port only gets instantiated at the first interaction with
> one of the Tomcats.
OK. :) How about BatCat ? It is stealthy, but still can say: "Miaú" :)
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
André,
On 4/7/2009 4:29 PM, André Warnier wrote:
> I would have to disagree again.
Sorry, I should have done the code in a haiku.
> The Tomcat you describe above is totally deterministic : it is alive
> (since it can receive your request), but as soon as you interact with
> it, it dies.
Hmm... you're right. I was thinking that the System.exit() would kill
the process and therefore the connection, but by the time the JVM dies,
the connection has already been made, proving the liveness of the
instance (nb: Mozilla Thunderbird does not consider 'liveness' to be a
correctly-spelled English word).
> Can [the listening port] be for example half-open ?
Perhaps. How about this:
ServerSocket socket = new ServerSocket(8080);
// No subsequent socket.accept() call
> And is a half-open port 8080 equivalent to a fully-open port 4040, or
> does it just have only half the bandwidth ?
Half the bandwidth, I think. "Half" would refer to its openness, not to
its port number, but maybe only half of the data gets through per unit
time. How many gigabits does the average cat hold? (nb: tb does not
consider 'gigabits' to be a correctly-spelled word, though 'gigabytes'
/is/).
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEUEARECAAYFAknbufcACgkQ9CaO5/Lv0PANUQCXd3QHBarYGT4YouC4saqc27xM
sgCgjILsEbGDO3+rKWZGx2aUZ7znzPY=
=EpK1
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by André Warnier <aw...@ice-sa.com>.
Christopher Schultz wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> André,
>
> On 4/7/2009 7:57 AM, André Warnier wrote:
>> A Schrödinger Tomcat would be one that exists multiple times, in quantum
>> superposition. Which would probably create problems with all of them
>> trying to listen on the same TCP ports. Unless of course the listening
>> port only gets instantiated at the first interaction with one of the
>> Tomcats.
>
> I would think that a Schrödinger Tomcat would be one configured thusly:
>
> 1. ROOT context captures all incoming requests and routes them to the
> CheckForTomcatRunningServlet.
> 2. CheckForTomcatRunningServlet is implemented as follows:
>
> public void service(...) {
> System.exit(0);
> }
>
> Tomcat is either up or not.
I would have to disagree again.
The Tomcat you describe above is totally deterministic : it is alive
(since it can receive your request), but as soon as you interact with
it, it dies.
In my view, the Schrödinger Tomcat's existence is never in doubt. There
are even an infinity of them, each one having some probability of being
more or less alive or dead. As soon as you interact with it/them
however, you automatically select one of them.
Thus the code above should at least incorporate the following
pseudo-code modification :
> public void service(...) {
if (random(1) < 0.5)
> System.exit(0);
else
response.print(locale("miaou"));
> }
The question in fact really boils down to whether these quantums of life
extend or not to the listening port. Can it be for example half-open ?
And is a half-open port 8080 equivalent to a fully-open port 4040, or
does it just have only half the bandwidth ?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
André,
On 4/7/2009 7:57 AM, André Warnier wrote:
> A Schrödinger Tomcat would be one that exists multiple times, in quantum
> superposition. Which would probably create problems with all of them
> trying to listen on the same TCP ports. Unless of course the listening
> port only gets instantiated at the first interaction with one of the
> Tomcats.
I would think that a Schrödinger Tomcat would be one configured thusly:
1. ROOT context captures all incoming requests and routes them to the
CheckForTomcatRunningServlet.
2. CheckForTomcatRunningServlet is implemented as follows:
public void service(...) {
System.exit(0);
}
Tomcat is either up or not. You are outside the "box" (ha ha ha) so you
can only check if it's up by sending an HTTP request to it. Is it running?
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAknboY4ACgkQ9CaO5/Lv0PCyiwCfUtPg9APi+7e5vQ7hKZHC1XWx
9/YAniya2ZZgU6Epe0R9spxHBeprz4s9
=QCAD
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by André Warnier <aw...@ice-sa.com>.
János Löbb wrote:
>
> On Apr 6, 2009, at 4:10 AM, André Warnier wrote:
>
>> Caldarale, Charles R wrote:
>> [...]
>>> The metaphysical implications of existing without a trace are rather
>>> intriguing...
>> I am surprised that you would not have heard of stealth technology.
>> What do they call a stealthy Tomcat ? a Raptor ?
> No.... that must be a Schrödingercat :)
I am sorry, but I will have to disagree.
A Schrödinger Tomcat would be one that exists multiple times, in quantum
superposition. Which would probably create problems with all of them
trying to listen on the same TCP ports. Unless of course the listening
port only gets instantiated at the first interaction with one of the
Tomcats.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by János Löbb <ja...@yale.edu>.
On Apr 6, 2009, at 4:10 AM, André Warnier wrote:
> Caldarale, Charles R wrote:
> [...]
>> The metaphysical implications of existing without a trace are
>> rather intriguing...
> I am surprised that you would not have heard of stealth technology.
> What do they call a stealthy Tomcat ? a Raptor ?
No.... that must be a Schrödingercat :)
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: [OT] Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by André Warnier <aw...@ice-sa.com>.
Caldarale, Charles R wrote:
[...]
>
> The metaphysical implications of existing without a trace are rather intriguing...
>
I am surprised that you would not have heard of stealth technology.
What do they call a stealthy Tomcat ? a Raptor ?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by Daniel Lopez <gr...@dev.java.net>.
Hi there,
2009/4/6 Caldarale, Charles R <Ch...@unisys.com>:
> The metaphysical implications of existing without a trace are rather intriguing...
Yeah, I was about to call Mulder and Scully to investigate it, but I
thought I'd give the user list a chance first.
> Looks like it. In Tomcat 5.5, there was a non-daemon thread for each <Connector> named "http-<port>-Monitor"; this thread no longer exists in 6.0, since the whole area appears to have been re-architected. There are no non-daemon threads in Tomcat 6.0, other than the whatever thread gets the game started; the normal Tomcat bootstrap uses this lead thread to listen on the shutdown port. You'll need to have some thread hanging around to keep the JVM from terminating; if you don't have such a thread, why are you bothering with embedded?
I usually ran the app with the GUI, but sometimes it is useful to ran
it without, specially during development and testing of new
containers, as I don't have to wait for the Windows to open. I did not
know it was "disallowed". OTOH, if I did not want my app to start
alive, why would I have started a server listening in a port? I use
embedded to test web applications and to run them inside Ant so I can
perform integration tests on them within Ant, so the GUI part is
optional.
>> if I start the server with a GUI window, so the container
>> remains alive due to the GUI thread, stopping the server
>> does not really work, as even though it says that the
>> Connector stopped, I can still access the server for some
>> seconds and then it starts answering "resource not available",
>> but the thing is, it still answers to the port.
>
> It's your responsibility to terminate your GUI thread so the JVM can kill off all the daemon threads.
Do you mean that the stop() method on the Server class is useless and
in order to stop the embedded server I have to kill my whole
application? That doesn't make much sense.
> You really use "this." everywhere? Seems a bit tedious, to say nothing of making code reviews difficult.
"this" is added automatically by the IDE and in this case, it helps me
distinguish between the objetcs that are created on the start method
and the ones I have to keep a reference to.
> You have to change your code; you got lucky in 5.5.
To me it makes more sense the 5.5 behaviour, but oh well :). In any
case, not being able to stop the server normally without killing my
whole application is really a non-no for me.
> Yes, the doc could be improved here. You could create a FAQ/Wiki entry based on your findings.
Once I get a clear picture, I'll try to.
> - Chuck
Thanks for your help,
D.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
RE: Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
Posted by "Caldarale, Charles R" <Ch...@unisys.com>.
> From: d.lopez.j@gmail.com [mailto:d.lopez.j@gmail.com] On Behalf Of
> Daniel Lopez
> Subject: Tomcat 5.5 embedded vs Tomcat 6.0.18 embedded
>
> when my other threads finish, the container simply exists
> without a trace.
The metaphysical implications of existing without a trace are rather intriguing...
> I've seen a thread in the tomcat-dev list ("embedded tomcat 6" 14 Feb
> 2008) that seems to imply this is normal with Tomcat 6, but in Tomcat
> 5.5 this did not happen. Is this normal?
Looks like it. In Tomcat 5.5, there was a non-daemon thread for each <Connector> named "http-<port>-Monitor"; this thread no longer exists in 6.0, since the whole area appears to have been re-architected. There are no non-daemon threads in Tomcat 6.0, other than the whatever thread gets the game started; the normal Tomcat bootstrap uses this lead thread to listen on the shutdown port. You'll need to have some thread hanging around to keep the JVM from terminating; if you don't have such a thread, why are you bothering with embedded?
> if I start the server with a GUI window, so the container
> remains alive due to the GUI thread, stopping the server
> does not really work, as even though it says that the
> Connector stopped, I can still access the server for some
> seconds and then it starts answering "resource not available",
> but the thing is, it still answers to the port.
It's your responsibility to terminate your GUI thread so the JVM can kill off all the daemon threads.
> The code looks like this:
> ----------------------------------
> this.theEmbedded = new Embedded();
> this.theEmbedded.setCatalinaHome(new File("tomcat_6").getAbsolutePath());
> // Add the engine
> this.engine = this.theEmbedded.createEngine();
> this.engine.setName("Catalina");
> this.engine.setDefaultHost("localhost");
> // Create the host
> this.host = this.theEmbedded.createHost("localhost", ".");
> this.engine.addChild(this.host);
> this.theContext = this.theEmbedded.createContext("/",
> this.theConfiguration.getBasePath());
> this.theContext.setReloadable(false);
> this.host.addChild(this.theContext);
> this.theEmbedded.addEngine(this.engine);
> // Add a conector
> this.connector =
> this.theEmbedded.createConnector((java.net.InetAddress) null,
> this.port, false);
> this.theEmbedded.addConnector(this.connector);
> this.theEmbedded.start();
You really use "this." everywhere? Seems a bit tedious, to say nothing of making code reviews difficult.
> So, is the Tomcat 6 embedded engine broken or do I have to
> change the code?
You have to change your code; you got lucky in 5.5.
> Nothing in the API seemed to indicate any changes needed
Yes, the doc could be improved here. You could create a FAQ/Wiki entry based on your findings.
- Chuck
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org