wrapper-user — User issues for Wrapper. (Support questions, etc.)

Re: [Wrapper-user] Starting and stopping multiple configured services

[<da...@ix...>]

In article <3ED...@ta...>,
Leif Mortenson  <wra...@li...> wrote:
>Looking into this, there does not seem to be an easy way to wait in a batch
>file.  But I did locate this page which describes how to use the PING
>command to ping a host for n seconds and then complete.  This will
>effectively pause your batch file for n seconds.

We used to use this method.

These days I recommend a java app with nothing more than Thread.sleep(X);
in it.  I mean, you're already requiring java anyway.  Might as well make
use of it.  :->

mrc

-- 
     Mike Castle      da...@ix...      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc

Re: [Wrapper-user] Starting and stopping multiple configured services

[Leif M. <le...@ta...>]

Leif Mortenson wrote:

> I'll do some more testing to make sure that "net stop" is not timing out
> because of a problem with the Wrapper.  The Wrapper should be notifying
> the Service Manager that it needs more time to shutdown cleanly every few
> moments.  Same happens as the server is starting up. 

I verified that the Wrapper is indeed correctly informing the
ServiceManager that the Wrapper needs more time to stop cleanly.  It
appears that the "net stop" command is just ignoring that and timing out
on its own.  Probably so the user on the command line doesn't think its
hung or something...  All well.

Good news though.  I think I came up with a solution to your problem.  See
if you think it is acceptable.
Rather than using "net stop" to stop the service, use the Wrapper to remove
the service.  The Wrapper will wait until the service has completely stopped
before it is removed, so this should give you the desired behavior.  Then
when you want to start the services back up, you will have to reinstall 
them.

To stop your services, you would use this:
---
C:\srv1\bin\Wrapper -r ..\conf\wrapper.conf
C:\srv2\bin\Wrapper -r ..\conf\wrapper.conf
C:\srv3\bin\Wrapper -r ..\conf\wrapper.conf
C:\srv4\bin\Wrapper -r ..\conf\wrapper.conf
C:\srv5\bin\Wrapper -r ..\conf\wrapper.conf
---

Then to start them back up, you would do this:
---
C:\srv5\bin\Wrapper -i ..\conf\wrapper.conf
net start srv5
C:\srv4\bin\Wrapper -i ..\conf\wrapper.conf
net start srv4
C:\srv3\bin\Wrapper -i ..\conf\wrapper.conf
net start srv3
C:\srv2\bin\Wrapper -i ..\conf\wrapper.conf
net start srv2
C:\srv1\bin\Wrapper -i ..\conf\wrapper.conf
net start srv1
---

Version 3.0.3 and below will display a message once per second saying
that the Wrapper is waiting for the service to stop.  I just reduced that
frequency to once every five seconds.  That will be in 3.0.4 when it is
released.

Let me know whether or not this will work for you.  Another possibility 
is to
create a simple tool that has the sole job of replacing the "net start"
command with something that does not time out.  But the above works as
is.

Cheers,
Leif

Re: [Wrapper-user] Starting and stopping multiple configured services

[Leif M. <le...@ta...>]

Paul Casanova wrote:

>Thanks for responding so fast!
>
>>I want to confirm that your problem is not that the Wrapper is timing
>>out and
>>killing the JVM prematurely.  I don't think it is from what you said.
>>    
>>
>I initially set it all up with DEBUG turned on, until I got it running with
>all errors gone.  I did have to increase the timeout periods, but yes I
>believe that when the services are stopped individually from the Services
>applet, they exit without error.
>
>You're right - even though "net stop" bombs out with an error ("service
>could not be stopped..." or something), the services do appear to have
>stopped in the Services applet.  I'm just a little nervous about putting
>that into production.
>
I'll do some more testing to make sure that "net stop" is not timing out
because of a problem with the Wrapper.  The Wrapper should be notifying
the Service Manager that it needs more time to shutdown cleanly every few
moments.  Same happens as the server is starting up.

>The extra method I was thinking of is for the nightly backup situation,
>where it's pretty much guaranteed that users will be off the system
>(midnight), perhaps another class could be written which ends the services
>more abruptly or something.  I'm pretty new on the services side of things,
>so this may either be a valid idea, or complete nonsense that I'm
>suggesting!!
>
That is possible by setting the shutdown timeouts to very short values 
within
the wrapper.conf file.  But there is only one of those.  The problem is that
there are not multiple commands coming from the ServiceManager based
on the time.

It would be possible to add something to the wrapper.conf where you could
define windows of time where the Wrapper would immediately kill the JVM
rather than requesting the JVM to shutdown cleanly.  If possible I would 
like
to avoid doing this as I don't think this a very common problem however.

You could implement this behavior easily in your own applications by
adding a check to the beginning of your application's shutdown hook.  If
the current time is within a certain window, execute the following code.
If you call halt outside a shutdown hook, and the Wrapper has not
requested that the JVM shutdown, then it will interpret the JVM exiting
as it having crashed and will restart the JVM.  The call to signalStopped
is to tell the Wrapper that the JVM is exiting intentionally so it is not
restarted.  System.exit will shutdown cleanly and the Wrapper will
handle it correctly.

WrapperManager.signalStopped( 0 )
Runtime.getRuntime().halt(0)

Hope this helps.

Cheers,
Leif

Re: [Wrapper-user] Starting and stopping multiple configured services

[Paul C. <cas...@au...>]

Thanks for responding so fast!

>I want to confirm that your problem is not that the Wrapper is timing
>out and
>killing the JVM prematurely.  I don't think it is from what you said.

I initially set it all up with DEBUG turned on, until I got it running with
all errors gone.  I did have to increase the timeout periods, but yes I
believe that when the services are stopped individually from the Services
applet, they exit without error.

You're right - even though "net stop" bombs out with an error ("service
could not be stopped..." or something), the services do appear to have
stopped in the Services applet.  I'm just a little nervous about putting
that into production.

The extra method I was thinking of is for the nightly backup situation,
where it's pretty much guaranteed that users will be off the system
(midnight), perhaps another class could be written which ends the services
more abruptly or something.  I'm pretty new on the services side of things,
so this may either be a valid idea, or complete nonsense that I'm
suggesting!!

I'll speak to management to see what can be done about supporting the Java
Service Wrapper project.

Thanks,

Paul Casanova

Re: [Wrapper-user] Starting and stopping multiple configured services

[Leif M. <le...@ta...>]

Paul Casanova wrote:

>First up Leif, I would like to thank and congratulate you and everyone else
>who has worked on the Java Service Wrapper for making such a fantastic
>configurable and scaleable product available to us all.
>
Thanks, glad you find it useful.

>I manage a complex Java application for a client.  The application has
>(now) 5 components as follows:
>Tomcat 4.1.24
>RMI Registry
>AutoUpdater
>MainServer (-Xmx1100)
>AdditionalServer (-Xmx1100)
>
Wow.  My limit so far has been 512MB

>All of these now start and stop beautifully individually via the Windows
>Services applet.  Previously when we took over support of the application,
>the "srvany" Microsoft tool (horrid) was being used to start and stop the
>components of the application individually and as a whole using "net
>stop..." and "net start..." commands inside batch scripts.
>
>I can start them all at once successfully from a batch file using "net
>start..." eg:
>net start service1
>net start service2
>...
>
>However, using this approach to stop all the services together (eg for
>nightly database backup) now times out though because the applications are
>being shut down cleanly, rather than just being killed off (ala "srvany").
>
>I have a dependency for all of the services (except Tomcat) on the
>RMIRegistry, which works fine through the Services applet - when it starts,
>they all start; when it stops, they all stop.  No problem.
>
>However, the "net stop" command ignores this dependency, and doesn't stop
>or start the other services when it is stopped or started.
>
>What is the preferred method of stopping several Java Service Wrapper
>services at the same time?  The first 3 services mentioned aren't a problem
>because they are relatively small and simple (< 20MB).  However, the last 2
>use up to 1.1GB of RAM each (yeah, I know - quite large) and have large
>database connection pools and things they need to clean up before they can
>exit cleanly.
>
I want to confirm that your problem is not that the Wrapper is timing 
out and
killing the JVM prematurely.  I don't think it is from what you said.

The only thing I can think of would be to either write a C (or Java) 
application
that would be a little smarter about waiting for each process to stop.

An easy way to do this would be to have your batch file pause for a 
minute or
so after each call to "net stop <apl>".  This way you could guarantee that
enough time was allowed for the application to actually stop.   I assume 
that
net stop is timing out, but that the Wrapper is continuing to stop the 
service
anyway and it will eventually exit on its own.

Looking into this, there does not seem to be an easy way to wait in a batch
file.  But I did locate this page which describes how to use the PING
command to ping a host for n seconds and then complete.  This will
effectively pause your batch file for n seconds.

http://www.robvanderwoude.com/index.html

Their example is pretty complex because it is designed to work on any
windows version.  But you basically just place a call like the following 
into
your batch file where you want to wait for 60 seconds:

PING 1.1.1.1 -n 60 -w 1000 >NUL

Just change the "60" to the number of seconds to wait.

>Is there any alternative to "net stop"?
>Can the wrapper be configured to have more than 1 shut down method?
>
What kind of additional shutdown method are you thinking would be useful?

If you found the product and support useful, please think about supporting
the project:  http://wrapper.tanukisoftware.org/doc/english/donate.html

Cheers,
Leif

[Wrapper-user] Starting and stopping multiple configured services

[Paul C. <cas...@au...>]

Hi,

This is my first post to this forum.

First up Leif, I would like to thank and congratulate you and everyone else
who has worked on the Java Service Wrapper for making such a fantastic
configurable and scaleable product available to us all.

I manage a complex Java application for a client.  The application has
(now) 5 components as follows:
Tomcat 4.1.24
RMI Registry
AutoUpdater
MainServer (-Xmx1100)
AdditionalServer (-Xmx1100)

All of these now start and stop beautifully individually via the Windows
Services applet.  Previously when we took over support of the application,
the "srvany" Microsoft tool (horrid) was being used to start and stop the
components of the application individually and as a whole using "net
stop..." and "net start..." commands inside batch scripts.

I can start them all at once successfully from a batch file using "net
start..." eg:
net start service1
net start service2
...

However, using this approach to stop all the services together (eg for
nightly database backup) now times out though because the applications are
being shut down cleanly, rather than just being killed off (ala "srvany").

I have a dependency for all of the services (except Tomcat) on the
RMIRegistry, which works fine through the Services applet - when it starts,
they all start; when it stops, they all stop.  No problem.

However, the "net stop" command ignores this dependency, and doesn't stop
or start the other services when it is stopped or started.

What is the preferred method of stopping several Java Service Wrapper
services at the same time?  The first 3 services mentioned aren't a problem
because they are relatively small and simple (< 20MB).  However, the last 2
use up to 1.1GB of RAM each (yeah, I know - quite large) and have large
database connection pools and things they need to clean up before they can
exit cleanly.

Is there any alternative to "net stop"?
Can the wrapper be configured to have more than 1 shut down method?

Any ideas on this would be much appreciated.

Thanks,

Paul Casanova
AMS Ballarat Application Centre
IBM Global Services Australia
Mount Helen

Telephone:
61-3-53273245
Fax:
61-3-53273297

[Wrapper-user] Re: Thread dump failure

[Sergey K. <ser...@ho...>]

Much appreciated!

Re: [Wrapper-user] Thread dump failure

[Leif M. <le...@ta...>]

Sorry about that.  I noticed it a couple weeks ago and have fixed it in the
CVS version.  It was a stupid typo in the C source. One line change.
It will be in the 3.0.4 release.
This new version also includes a couple of bug fixes with the sh and bash
scripts used to control the Wrapper.

Could you check out the latest from CVS?  It will fix your problems.  I'll
try to get a new version released fairly soon, but there are a few more
things that I want to get into the next release.

Cheers,
Leif

Sergey Kapustin wrote:

> Hello,
> I'm trying out the wrapper and ran into one problem. When I request a 
> thread dump through the shell script first time, everything works as 
> expected. Although, when I try to get dump for the second time, the 
> wrapper simply kills the JVM.
> wrapper version is 3.0.3, it was tested on Solaris 5.8 and RedHat 
> Linux 7.2
> Could you please explain why the wrapper is doing it and how to fix it?
>  
> Thank you
>  
> Here is the wrapper log with normal dump:
> --------------------------------------------------
> wrapper  | Dumping JVM state.
> jvm 1      | Full thread dump:
>          .....
>  
> This what happens when dump is requested for the second time:
> --------------------------------------------------
> wrapper  | Shutting down.
> wrapper  | wrapperStopProcess(0) called.
> wrapper  | Sending stop signal to JVM
> wrapperp | send a packet 101 : NULL
> jvm 1    | Received a packet 101 :
> jvm 1    | Thread, Wrapper-Connection, handling the shutdown process.
> jvm 1    | calling listener.stop()
> jvm 1    | WrapperSimpleApp: stop(0)
> jvm 1    | returned from listener.stop()
> jvm 1    | Send a packet 107 : 0
> jvm 1    | Closing socket.
> wrapperp | read a packet 107 : 0
> wrapper  | JVM signalled that it was stopped.
> wrapperp | socket read no code (closed?).
> jvm 1    | calling System.exit(0)
> wrapper  | JVM exited normally.
> wrapper  | <-- Wrapper Stopped
>  

[Wrapper-user] Thread dump failure

[Sergey K. <ser...@ho...>]

Hello,
I'm trying out the wrapper and ran into one problem. When I request a =
thread dump through the shell script first time, everything works as =
expected. Although, when I try to get dump for the second time, the =
wrapper simply kills the JVM.=20
wrapper version is 3.0.3, it was tested on Solaris 5.8 and RedHat Linux =
7.2
Could you please explain why the wrapper is doing it and how to fix it?

Thank you

Here is the wrapper log with normal dump:
--------------------------------------------------
wrapper  | Dumping JVM state.
jvm 1      | Full thread dump:
         .....

This what happens when dump is requested for the second time:
--------------------------------------------------
wrapper  | Shutting down.
wrapper  | wrapperStopProcess(0) called.
wrapper  | Sending stop signal to JVM
wrapperp | send a packet 101 : NULL
jvm 1    | Received a packet 101 :=20
jvm 1    | Thread, Wrapper-Connection, handling the shutdown process.
jvm 1    | calling listener.stop()
jvm 1    | WrapperSimpleApp: stop(0)
jvm 1    | returned from listener.stop()
jvm 1    | Send a packet 107 : 0
jvm 1    | Closing socket.
wrapperp | read a packet 107 : 0
wrapper  | JVM signalled that it was stopped.
wrapperp | socket read no code (closed?).
jvm 1    | calling System.exit(0)
wrapper  | JVM exited normally.
wrapper  | <-- Wrapper Stopped

[Wrapper-user] Re: Intended user id under Unix daemon

[<da...@ix...>]

In article <D0C...@co...>,
Higgins, Steven C (Steven) <wra...@li...> wrote:
>I can easily modify the JSW wrapper script to start JBoss using the 'su'
>command and specify my desired user with the '-l' option.  However, I
>don't think that was the intent.

This is precisely the method that we intend to use.

mrc
-- 
     Mike Castle      da...@ix...      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc

[Wrapper-user] Intended user id under Unix daemon

[Higgins, S. C (Steven) <sch...@av...>]

Under what user id is the wrapper intended to run as a daemon?

Without modification I can run everything as root.

I'm running JBoss and would like to run everything as a specific user =
id.

I can easily modify the JSW wrapper script to start JBoss using the 'su' =
command and specify my desired user with the '-l' option.  However, I =
don't think that was the intent.

What solutions have you used to address this? What is the preferred =
solution?

Just looking for ideas!

Thanks,
Steven Higgins

Re: [Wrapper-user] set environment path variable (win32)

[Leif M. <le...@ta...>]

Mat,
    What version of the Wrapper are you using?  Evaluation of environment
variables was added in version 2.2.9.   Try enabling debug output and look
at the command used to launch the JVM, you can then see whether the
problem is with the wrapper.conf file or on the Java side.

    I placed the following in the wrapper.conf:
wrapper.java.additional.1=-DPATH="%PATH%;C:\Test\bin;"
    and the following in a Java class:
System.out.println( "Path=" + System.getProperty( "PATH" ) );
    and things worked fine.

    Check the debug output, that will narrow down the problem.  You can
enable all debug output by setting
wrapper.debug=true

Cheers,
Leif

M. Leinmueller wrote:

> Leif,
>
> If I interpret your comments and the documentation correctly something 
> like
>   wrapper.java.additional.1=-DPATH="%PATH%;C:\Test\bin;"
> should take the current environment path, add  "C:\Test\bin" and give 
> the whole thing as path parameter to the JVM.
> This does not work for me :-(
>
> Mat

Re: [Wrapper-user] set environment path variable (win32)

[M. L. <m_l...@ho...>]

Leif,

If I interpret your comments and the documentation correctly something like
   wrapper.java.additional.1=-DPATH="%PATH%;C:\Test\bin;"
should take the current environment path, add  "C:\Test\bin" and give the 
whole thing as path parameter to the JVM.
This does not work for me :-(

Mat

>
>Mat,
>    You can set any environment variable from within the wrapper.conf file 
>by
>creating a property with the following syntax:
>set.PATH=...your path...
>
>    See the following page for more details.
>http://wrapper.tanukisoftware.org/doc/english/props-envvars.html
>
>    What are you trying to do though?  The JVM is not able to access
>environment variables directly.  You normally pass environment data to
>a JVM by defining properties on the Java command line:
>
>java -Dpath=%PATH% ...
>
>    If your JVM is spawning other processes, then they would have access
>to the environment variables set in the wrapper.conf file.
>
>    If you find the product and support useful, please think about 
>supporting
>the project:  http://wrapper.tanukisoftware.org/doc/english/donate.html
>
>Cheers,
>Leif

_________________________________________________________________
MSN Hotmail  -  Absolut kostenfrei! Der weltweit größte E-Mail-Anbieter im 
Netz:  http://www.msn.de/hotmail

Re: [Wrapper-user] set environment path variable (win32)

[Leif M. <le...@ta...>]

Mat,
    You can set any environment variable from within the wrapper.conf 
file by
creating a property with the following syntax:
set.PATH=...your path...

    See the following page for more details.
http://wrapper.tanukisoftware.org/doc/english/props-envvars.html

    What are you trying to do though?  The JVM is not able to access
environment variables directly.  You normally pass environment data to
a JVM by defining properties on the Java command line:

java -Dpath=%PATH% ...

    If your JVM is spawning other processes, then they would have access
to the environment variables set in the wrapper.conf file.

    If you find the product and support useful, please think about 
supporting
the project:  http://wrapper.tanukisoftware.org/doc/english/donate.html

Cheers,
Leif

M. Leinmueller wrote:

> Hi,
>
> is it possible to set the environment PATH variable (win32) used by 
> the JVM for a WrapperSimpleApp?
> E. g. in my batch file I set PATH="%PATH%;..\lib", how can I do this 
> with Java Service Wrapper?
>
> Thanks in advance.
>
> Mat 

[Wrapper-user] set environment path variable (win32)

[M. L. <m_l...@ho...>]

Hi,

is it possible to set the environment PATH variable (win32) used by the JVM 
for a WrapperSimpleApp?
E. g. in my batch file I set PATH="%PATH%;..\lib", how can I do this with 
Java Service Wrapper?

Thanks in advance.

Mat

_________________________________________________________________
Mit dem MSN Messenger eine Reise für 4 Personen nach Barcelona gewinnen – 
jetzt mitmachen! http://www.sweepstakes2003.com/entry.aspx?locationid=15

Re: [Wrapper-user] An option in wrapper.conf to configure the startWaitHint?

[Leif M. <le...@ta...>]

Jindong,
    I will take a look at how to improve this with the WrapperSimpleApp and
WrapperStartStopApp helper classes.  As is described in the documentation,
this is expected.  But it can use improvement.
    For now, to get this behavior, you can use integration method 3 and
implement the WrapperListener interface yourself.  This method requires
more work however so getting your request implemented will be the best
solution.

Cheers,
Leif

Jindong Li wrote:

> Hi there,
>
> I'm using WrapperStartStopApp to start my application as Windows 
> Service, one thing I think would be a nice to have is to be able to 
> set up the startWaitHint in the configuration file, right now, the 
> start() will only wait 2 seconds before it returns and signaling that 
> the service has started while in fact the thread that runs the main 
> method could be still running especially with applications that take a 
> long time to start...
>
> This will avoid the situation where service appears started even 
> though it's not...
>
> Any comments?
>
> Jindong.
>

Re: SOLVED RE: [Wrapper-user] Environment problems?

[Leif M. <le...@ta...>]

Eric,
Glad you got things working.  Let me know if you come across any other
problems.

If you found the product and support useful, please think about supporting
the project:  http://wrapper.tanukisoftware.org/doc/english/donate.html

Cheers,
Leif

Roberts, Eric wrote:

>Leif,
>
>I unintentionally ran this program from the command line without having run .profile and got exactly the same error - so the problem is the environment variables, and nothing to do with the wrapper.
>
>Someone has screwed up the .profile script on my test system!
>
>Thanks for your clarification about the environment variables anyway.
>
>Regards
>
>Eric 
>  
>

Re: [Wrapper-user] Environment problems?

[Leif M. <le...@ta...>]

Eric,
The Wrapper opens a single socket that accepts connections from the JVM
when the WrapperManager class initializes.  This socket is then used for
all communication between the Wrapper and the JVM.  You might want to
try changing the value of wrapper.port to another value and see if that
changes anything.   It is a TCP connection, and your bus transport  uses
UDP, so I doubt that is the problem.

The next thing I would suggest trying is to copy the entire command used
to launch the Java process into a shell script and try running the exact
same command.  This is an easy way to eliminate the Wrapper as the
cause of your problems.   It will still use the WrapperManager within the
JVM but in my experience, there has never been a problem that I heard
about caused by that class being loaded.
You do need to make one change to the command line from that shown
in the debug output of the wrapper.  remove the wrapper.key property.
If it is defined, then the WrapperManager class will think it is supposed
to be communicating with the Wrapper and will exit the JVM when it fails
to do so.  Otherwise there should be no changes.

I will be interested to see if your application works or fails there.

It is possible that your application may be loading native libraries.
Normally most java applications do not specify a java library path, so
the JVM will use the default.  In the case of the wrapper, you must specify
the location of the libwrapper.so file on in the wrapper.conf file.  This
means that the default library path will not be searched.  To fix this you
would need to add the location of the additional libraries as additional
library path entries in the wrapper.conf file.  If this is the problem, then
the above test should fail.

Cheers,
Leif


Roberts, Eric wrote:

>Leif,
>
>Firstly thanks for the response.
>
>The application is third party (Tibco Integration Manager), so I can only give you a rough idea of what is taking place.
>
>The program is designed to be fault tolerant - the first "engine" is started with a flag which tells it to establish a fault tolerant bus transport, subsequent "engines" search for existing engines with the same group name and form a fault tolerant cluster. The exception is occurring when the first instance attempts to establish the fault tolerant bus transport (hence the error message "Initialization failed for bus transport FtDefault"). I believe that this transport is UDP.
>
>Because I can successfully start the program from the command line, I thought at first to ask about how the wrapper passed environment settings. From your response it would appear that the environment is not the issue.
>
>Do you think that there could be some system resource conflict between the wrapper, and the application trying to establish the transport?
>
>Another thought was permissions - but again, if I can run the program successfully from the same shell as I try to run the wrapper, then I cannot see that this could be an issue.
>
>I hope that some genius out there might be able to hint at what the problem might be and a possible fix, as I currently have many of these programs to support in a production environment, and the wrapper seemed the perfect solution for automatic restarting.
>
>Regards
>
>Eric
>  
>

SOLVED RE: [Wrapper-user] Environment problems?

[Roberts, E. <Eri...@on...>]

Leif,

I unintentionally ran this program from the command line without having =
run .profile and got exactly the same error - so the problem is the =
environment variables, and nothing to do with the wrapper.

Someone has screwed up the .profile script on my test system!

Thanks for your clarification about the environment variables anyway.

Regards

Eric=20

-----Original Message-----
From: Leif Mortenson [mailto:le...@ta...]
Sent: 21 May 2003 18:21
To: wra...@li...
Subject: Re: [Wrapper-user] Environment problems?


Eric,
    Do you know what part of your program is kicking this error out?  It
appears as if the Wrapper launches the WrapperSimpleApp helper
normally and calls your configured main method.
    The main method then displays some info about the current JVM,
Displays an error and then it looks like it calls System.exit()  I say =
that
because during the same second, the shutdown hook is triggered.

>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: start(args)
>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: invoking =
main method
>INFO   | jvm 1    | 2003/05/19 13:32:58 | Using Java HotSpot(TM) Client =
VM, 1.3.1-b24, mixed mode
>INFO   | jvm 1    | 2003/05/19 13:32:59 | exception: Initialization =
failed for bus transport FtDefault
>INFO   | jvm 1    | 2003/05/19 13:32:59 | Wrapper Manager: ShutdownHook =
started
> =20
>
    Without knowing what you program is doing here is is difficult to
help. I tried doing a search but came up with nothing.  If you could =
explain
what your code is doing there, it might help one of us to come up with
some ideas.

    As fo the environment, that is entirely possible.  The Wrapper =
launches
the JVM using the same user as was used to launch the Wrapper process.
The entire environment available to the Wrapper should also be available
to the JVM.   Most environment related problems tend to be with the
environment in which the Wrapper itself is launched.  I have not heard =
of
any problems, nor should there be any, where the JVM does not have
access to some part of the Wrapper's environment.  The JVM process
is just forked, so the two should be identical.

Cheers,
Leif




-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
Wrapper-user mailing list
Wra...@li...
https://lists.sourceforge.net/lists/listinfo/wrapper-user

RE: [Wrapper-user] Environment problems?

[Roberts, E. <Eri...@on...>]

Leif,

Firstly thanks for the response.

The application is third party (Tibco Integration Manager), so I can =
only give you a rough idea of what is taking place.

The program is designed to be fault tolerant - the first "engine" is =
started with a flag which tells it to establish a fault tolerant bus =
transport, subsequent "engines" search for existing engines with the =
same group name and form a fault tolerant cluster. The exception is =
occurring when the first instance attempts to establish the fault =
tolerant bus transport (hence the error message "Initialization failed =
for bus transport FtDefault"). I believe that this transport is UDP.

Because I can successfully start the program from the command line, I =
thought at first to ask about how the wrapper passed environment =
settings. From your response it would appear that the environment is not =
the issue.

Do you think that there could be some system resource conflict between =
the wrapper, and the application trying to establish the transport?

Another thought was permissions - but again, if I can run the program =
successfully from the same shell as I try to run the wrapper, then I =
cannot see that this could be an issue.

I hope that some genius out there might be able to hint at what the =
problem might be and a possible fix, as I currently have many of these =
programs to support in a production environment, and the wrapper seemed =
the perfect solution for automatic restarting.

Regards

Eric

-----Original Message-----
From: Leif Mortenson [mailto:le...@ta...]
Sent: 21 May 2003 18:21
To: wra...@li...
Subject: Re: [Wrapper-user] Environment problems?


Eric,
    Do you know what part of your program is kicking this error out?  It
appears as if the Wrapper launches the WrapperSimpleApp helper
normally and calls your configured main method.
    The main method then displays some info about the current JVM,
Displays an error and then it looks like it calls System.exit()  I say =
that
because during the same second, the shutdown hook is triggered.

>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: start(args)
>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: invoking =
main method
>INFO   | jvm 1    | 2003/05/19 13:32:58 | Using Java HotSpot(TM) Client =
VM, 1.3.1-b24, mixed mode
>INFO   | jvm 1    | 2003/05/19 13:32:59 | exception: Initialization =
failed for bus transport FtDefault
>INFO   | jvm 1    | 2003/05/19 13:32:59 | Wrapper Manager: ShutdownHook =
started
> =20
>
    Without knowing what you program is doing here is is difficult to
help. I tried doing a search but came up with nothing.  If you could =
explain
what your code is doing there, it might help one of us to come up with
some ideas.

    As fo the environment, that is entirely possible.  The Wrapper =
launches
the JVM using the same user as was used to launch the Wrapper process.
The entire environment available to the Wrapper should also be available
to the JVM.   Most environment related problems tend to be with the
environment in which the Wrapper itself is launched.  I have not heard =
of
any problems, nor should there be any, where the JVM does not have
access to some part of the Wrapper's environment.  The JVM process
is just forked, so the two should be identical.

Cheers,
Leif




-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
Wrapper-user mailing list
Wra...@li...
https://lists.sourceforge.net/lists/listinfo/wrapper-user

Re: [Wrapper-user] Wrapper and Java 1.4.X

[<PB...@ba...>]

Hey

I upgraded to Wrapper version 3.0.3 and now it all works with the 1.4.2
beta JVM

Thanks for the help !

Peer




|---------+---------------------------------------->
|         |           Leif Mortenson               |
|         |           <le...@ta...>    |
|         |           Sent by:                     |
|         |           wra...@li...|
|         |           ceforge.net                  |
|         |                                        |
|         |                                        |
|         |           22-05-2003 13:27             |
|         |           Please respond to            |
|         |           wrapper-user                 |
|         |                                        |
|---------+---------------------------------------->
  >------------------------------------------------------------------------------------------------------|
  |                                                                                                      |
  |       To:       wra...@li...                                                   |
  |       cc:                                                                                            |
  |       Subject:  Re: [Wrapper-user] Wrapper and Java 1.4.X                                            |
  >------------------------------------------------------------------------------------------------------|




Mike Castle wrote:

>In answer to the original question though, I use wrapper with all versions
>of 1.4.1 (depends on what the developer has on their machine :-).
>
>Not tried the 1.4.2 stuff yet (I'm surprised to see a cutting edge JVM,
but
>not a cutting edge Wrapper).
>
Oh, I missed that. My brain stopped at 1.4.x I have not yet tested the
Wrapper with the 1.4.2 JVM. I'll have to give that a try.

Cheers,
Leif




-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
Wrapper-user mailing list
Wra...@li...
https://lists.sourceforge.net/lists/listinfo/wrapper-user

Re: [Wrapper-user] Wrapper and Java 1.4.X

[Leif M. <le...@ta...>]

Mike Castle wrote:

>In answer to the original question though, I use wrapper with all versions
>of 1.4.1 (depends on what the developer has on their machine :-).
>
>Not tried the 1.4.2 stuff yet (I'm surprised to see a cutting edge JVM, but
>not a cutting edge Wrapper).
>
Oh, I missed that. My brain stopped at 1.4.x I have not yet tested the
Wrapper with the 1.4.2 JVM. I'll have to give that a try.

Cheers,
Leif

Re: [Wrapper-user] Wrapper and Java 1.4.X

[<da...@ix...>]

In article <3EC...@ta...>,
Leif Mortenson  <wra...@li...> wrote:
>I would start by upgrading to version 3.0.3 and then rereading the new
>integration guide.  It was not very clear in the version you are trying to
>use.  There have been a lot of other improvements as well that would
>make it worth your while.


In answer to the original question though, I use wrapper with all versions
of 1.4.1 (depends on what the developer has on their machine :-).

Not tried the 1.4.2 stuff yet (I'm surprised to see a cutting edge JVM, but
not a cutting edge Wrapper).

mrc

-- 
     Mike Castle      da...@ix...      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc

[Wrapper-user] An option in wrapper.conf to configure the startWaitHint?

[Jindong Li <Jin...@so...>]

Hi there,

I'm using WrapperStartStopApp to start my application as Windows Service,
one thing I think would be a nice to have is to be able to set up the
startWaitHint in the configuration file, right now, the start() will only
wait 2 seconds before it returns and signaling that the service has started
while in fact the thread that runs the main method could be still running
especially with applications that take a long time to start...

This will avoid the situation where service appears started even though it's
not...

Any comments?

Jindong.

 
###############################################################
This message has been scanned for viruses and other contents.
###############################################################

Re: [Wrapper-user] Environment problems?

[Leif M. <le...@ta...>]

Eric,
    Do you know what part of your program is kicking this error out?  It
appears as if the Wrapper launches the WrapperSimpleApp helper
normally and calls your configured main method.
    The main method then displays some info about the current JVM,
Displays an error and then it looks like it calls System.exit()  I say that
because during the same second, the shutdown hook is triggered.

>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: start(args)
>INFO   | jvm 1    | 2003/05/19 13:32:58 | WrapperSimpleApp: invoking main method
>INFO   | jvm 1    | 2003/05/19 13:32:58 | Using Java HotSpot(TM) Client VM, 1.3.1-b24, mixed mode
>INFO   | jvm 1    | 2003/05/19 13:32:59 | exception: Initialization failed for bus transport FtDefault
>INFO   | jvm 1    | 2003/05/19 13:32:59 | Wrapper Manager: ShutdownHook started
>  
>
    Without knowing what you program is doing here is is difficult to
help. I tried doing a search but came up with nothing.  If you could explain
what your code is doing there, it might help one of us to come up with
some ideas.

    As fo the environment, that is entirely possible.  The Wrapper launches
the JVM using the same user as was used to launch the Wrapper process.
The entire environment available to the Wrapper should also be available
to the JVM.   Most environment related problems tend to be with the
environment in which the Wrapper itself is launched.  I have not heard of
any problems, nor should there be any, where the JVM does not have
access to some part of the Wrapper's environment.  The JVM process
is just forked, so the two should be identical.

Cheers,
Leif