david, Could we modify this message to say that - ERROR [GBeanInstance] GBeanInstance should already be stopped before die() is called. This GBean did not start due to errors. objectName=...................... Thanks Anita
--- anita kulshreshtha <[EMAIL PROTECTED]> wrote: > > > --- David Jencks <[EMAIL PROTECTED]> wrote: > > > > > On Jan 29, 2006, at 7:29 AM, anita kulshreshtha > > wrote: > > > > > David, > > > I was recently getting this error because I > > was > > > trying to get a reference for a nonexistent > Gbean > > (bad > > > program). Other Gbeans were waiting for this one > > to > > > start, I got a warning saying some beans were > not > > > started during server startup. During shutdown > > these > > > same beans (state starting) gave this error > > message. > > > Is it possible that the deployer is not flagging > > this > > > message? Is it something you want me to look > into > > and > > > where should I start? Is it just a case of > cryptic > > > error meassage? > > > > I don't really understand yet what is a problem > > here. To me it seems > > like everything is working properly. Perhaps > there > > can be > > improvements, but let's be sure we understand the > > situation before we > > start changing things. > > > > Let me try to explain what I think is typically > > happening and why it > > is difficult to avoid. > > > > Whenever a gbean is loaded and cant be started due > > to an unsatisfied > > reference, we log a notice indicating the gbean > and > > the reference > > pattern(s). > > > > When starting the server, after all configurations > > listed in > > config.xml are loaded, we log at a higher level > any > > gbeans that > > remain unstarted. > In my case, I got a warning on the console > saying > "some Gbeans were not started". When the above > message > "GBeanInstance ....." appeared on the console, I > knew > what it was due to. I looked at the logs and found > that these were the Gbeans that were not started. In > Nelson's case he is not getting any warning on the > console. I was just wondering why? > > Thanks > Anita > > > > When shutting down the server, we try to stop and > > kill (? not sure > > what it's called) all gbeans, and this results in > > logging a message > > for gbeans that were never started. > > > > Things we could do, and possible consequences: > > > > 1. when shutting down the server, terminate > > unstarted gbeans in such > > a way as to avoid logging the message about how > they > > can't be shut > > down. This won't produce any repercussions, but > > currently this > > annoying message is often the only visible way to > > find out that a > > gbean didn't fully start (it is difficult to read > > through so many > > debug log statements if you don't know you need to > > :-) > > > > 2. either log a visible message whenever a > > configuration starts but > > one or more gbeans can't start, or actually shut > > down the > > configuration in this case. This has some effects > > of discouraging or > > prohibiting some possible gbean relationships. > The > > simplest is where > > there are 2 configurations A and B: B uses classes > > from A but a gbean > > in A has a reference to a gbean in B. In this > case > > we have to load A > > first so thee classes for B are available, but the > > gbean in A can't > > start until after B has started. We actually have > > this situation > > today with j2ee-deployer and client-deployer. > (also > > openejb- > > deployer). A related problem occurs if there are > > configurations C > > and D with unrelated classloaders but C has gbeans > j > > and l and D a > > gbean k and there are references j >> k >> l. I'm > > hesitant to make > > these kinds of relationships noisy by logging a > > visible message or > > prohibiting them. > > > > So, I don't see any change that will produce a > > definite benefit. I'm > > certainly open to discussion. I wonder if we > should > > have this > > discussion on the dev list, you can certainly move > > my comments there. > > > > thanks > > david jencks > > > > > > > > Thanks > > > Anita > > > > > > --- "Nelson A. Perez" <[EMAIL PROTECTED]> > wrote: > > > > > >> > > >> Hi All, > > >> > > >> Thanks David Jencks for your comment on my > > >> previous > > >> post, but yes, I meant to write in the code I > > >> posted: > > >> > > >> > > >> ... > > >> <gbean name="gbean1" > > class="ibmtask.APrinterGBean"> > > >> </gbean> > > >> ... > > >> <reference name="APrinter"> <name>gbean1</name> > > >> </reference> > > >> ... > > >> > > >> That's the way my config plan looks like. I > wrote > > it > > >> the other way by accident ;-) Still, I can not > > make > > >> my > > >> application work properly. This is the error > that > > >> gets > > >> written into var/log/geronimo.txt when I do the > > >> deployment of my sample application: > > >> > > >> ERROR [GBeanInstance] GBeanInstance should > > already > > >> be > > >> stopped before die() is called: > > >> > objectName=geronimo.server:J2EEApplication=null, > > >> > > > J2EEModule=ibmtask/ > > > > > > IBMTask,J2EEServer=geronimo,j2eeType=GBean,name=MyGBean > > >> state=stopping > > >> > > >> Any suggestion on how to fix this issue ? > > >> > > >> Thanks, > > >> NP. > > >> > > >> > > __________________________________________________ > > >> Do You Yahoo!? > > >> Tired of spam? Yahoo! Mail has the best spam > > >> protection around > > >> http://mail.yahoo.com > > >> > > > > > > > > > > __________________________________________________ > === message truncated === __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
