Regardless...AFAICT...the error is gone in Tomcat. So whatever folks did in the last 24 hours fixed this problem (or so it seems). I now have 4 confirmed that its running w/o the warning...but I won't claim victory yet. Lets see if others get it.
The only problem we get now is an INFO message on shutdown of the root context saying that the context has already been shut down: 23:38:07,981 INFO [StandardContext] Container org.apache.catalina.core.ContainerBase.[Geronimo].[localhost].[/] has not been started This is a harmless message (not a warning or error). This is a life cycle issue as it looks like some other component already shut the context down...so this has no negative effect other than its ugly. I will try to track down who/what is shutting this down early. Jeff Dain Sundstrom wrote: > I don't think this was a Tomcat issue other then the race only occurred > in Tomcat. > > -dain > > On Dec 17, 2005, at 9:01 AM, Jeff Genender wrote: > >> I will have a look and see if this was related to the Tomcat issue... >> >> Thanks. >> >> Dain Sundstrom wrote: >>> I fixed the race condition in the day trader initialization, and >>> committed it to OpenEJB. The problem was cause by the web application >>> reveiving the notification that the TradeEJB (or another one) was >>> started before the ContainerIndex received the notification. When the >>> web application attempted to look up the EJB in the container index it >>> was not available. I coded around this problem by having the container >>> index check the kernel directly before failing the lookup, but we need a >>> more general solution to address the non-deterministic notification >>> ordering. >>> >>> -dain >>> >>> On Dec 16, 2005, at 11:48 AM, Matt Hogstrom wrote: >>> >>>> All, >>>> >>>> Rather than continue on one of the outstanding release chains its >>>> appropriate to start a new thread on where we are at and what we are >>>> doing. >>>> >>>> First, the outstanding issues with the release candidates clearly show >>>> that there are some issues that need to be resolved. The most >>>> significant is the error that is thrown during intialization. It is >>>> not, as was previously thought, an issue with copying the database >>>> files upon initial execution. It is rather associated with a race >>>> condition between the web-app being initialized and the EJB container >>>> reaching an initialized state. We are working on this issue and as >>>> such all previous release candidates should be discarded and ignored >>>> for now. We will track down the issues related to initialization as >>>> well as the other problems reported and put out a new release >>>> candidate. >>>> >>>> Second, thanks to all that are doing the testing and taking time to >>>> download and install the images on various operating systems. We very >>>> much appreciate your investement of time and interest. Your feedback >>>> is noted and will be incorporated where possible. >>>> >>>> Here are the outstanding issue that I've noted from the feedback on >>>> the list: >>>> >>>> GERONIMO-1363 - DayTrader still using old geronimo-spec files - fixed >>>> GERONIMO-1371 - Geronimo startup/shutdown issues >>>> GERONIMO-1372 - Exception during startup - TradeEJB >>>> GERONIMO-1375 - Invalid login to console should not produce stack trace >>>> >>>> If I missed one please let me know. >>>> >>>> Regarding the press release I'd like to ask Bruce and Geir to tackle >>>> that issue. Can you guys help out there? >>>> >>>> Thanks. >>>> >>>> Matt
