On Dec 17, 2005, at 1:14 PM, Jeff Genender wrote:
The Engine is starting before the host.
After chatting with Dain...the problem seems to be...
The plan is looking for a collection of hosts. But this does not seem
to place the dependency when there is a collection, only when its a
single GBean reference. Therefore its random of who gets started
first...call it luck of the draw? Therefore, this can rear its
head in
HEAD (no pun intended).
The problem stems from the fact that the user can have "n" number of
hosts...and the Engine needs to wait until all of these are started.
I would have hoped that the Collection reference would act like a
direct
GBean reference and wait until all have started, but this was my
misunderstanding. Can this be a problem for other GBeans that have
collection dependencies down the road - not just for Tomcat?
IMHO, I think the fact the Engine is starting before the Hosts is
dangerous and very bad.
I'll work on a solution of some form...
The only possibility I've thought of so far is a linked list starting
with the Engine with a reference to the first or last Host, and each
Host with a reference to another. I use something like this to order
the servlet startup in jetty.
david jencks
Jeff
Dave Colasurdo wrote:
Does anybody know why the "unknown default host" warning still
occurs in
1.0 but not in head? I thought Dain's fix was picked up by both
1.0 and
head? I also thought the next 1.0 candidate build would be made from
the 1.0 branch and not head? Is this incorrect?
Thanks
-Dave-
Jeff Genender wrote:
I was too darn quick with my send button...
Yes...let me follow up...
Building form HEAD...this error (1377) does not occur...I should
have
made that more clear in all of my previous emails.
Dave Colasurdo wrote:
I've extracted and built a fresh copy of the 1.0 branch..
I no longer see the TradeEJB exception reported in
GERONIMO-1372!!!!!!
However, I still see the "unknown default host" reported in
GERONIMO-1377. So, doesn't look like the 1372 fix addresses the
1377
problem. Jeff had suggested building head. Is there some other
fix in
head that isn't in the 1.0 branch that addresses this? Last time I
built head (last night).. I didn't see the "unknown default
host" error.
Also, concerning the conversation as to whether 1377 is a
showstopper..
Of course I'm sure that Jeff's assessment that there is no
functional
problem is totally accurate. However, I suspect the presence of
this
message in the log at initial startup will cause confusion and many
questions among new v1 users. It may give the wrong initial
perception
that the default geronimo v1 configuration doesn't work properly.
Hopefully Jeff will tell me that this is fixed in "head" and will
somehow be moved to v1..
On an unrelated note, I did see an exception during shutdown ..
http://www.rafb.net/paste/results/H4vwRo78.html
Is this covered under any other JIRA?
BTW, How long do the www.rafb.net paste entries last?
Thanks
-Dave-
Matt Hogstrom wrote:
Here is the lost of outstanding defects and their status:
In Process
GERONIMO-1364 - update welcome pages to point at HTTP redirects
in the
geronimo.apache.org site
GERONIMO-1371 - Geronimo startup/shutdown issues
GERONIMO-1375 - Invalid login to console should not produce
stack trace
GERONIMO-1377 - Startup Warning on tomcat - unknown default host.
Completed
GERONIMO-1363 - DayTrader still using old geronimo-spec files -
applied by Matt
GERONIMO-1372 - Exception during startup - TradeEJB - Fixed per
Dain
GERONIMO-1373 - DB info portlet not working correctly - Applied
by Matt
I would like to get the reminaing 4 knocked out today and respin a
build tonight if all these issues are addressed or we know they
will
not be. The major ones are behind us. It sounds like we're
going to
have to spin an Open EJB 2.1.0 or some such release per the fix
from
Dain.
Any volunteers on the other issues? Please assign them to
yourself or
send a note in this thread with your intent so everyone knows
what the
others are doing.
Great work folks. I think we're almost there.
I have to run some Christmas shopping errands this morning
(EST) and
will get back on in the afternoon. If someone needs me for some
reason please feel free to call my cell (919)656-0564.
Cheers.
Matt
