Hello Elena
2009/11/24 Elena Asarina <Elena.Asarina_at_sun.com>
> Based on my experience everything depends from the archive size. I have
> about 5 archives that have a size around 5 MB. My test executes with each
> archive 20 actions like deploy/undeploy/redeploy and so on. If to execute
> with these 5 archive that 100 actions totally, then with a big probability
> appserver either hung up, for example will occupy 100% CPU, or will crash.
> Early in a testing cycle I've opened bugs that describe these issues,
> then eventually these bugs were closed.
Seems like this will only be fixed in GlassFish 3.1 sorry unless we can
motivate nicely.
I've seen in older version of JSF. i.e. pre 2.0 release the problem is more
apparent. So make sure you use 2.0.1 + or the one in the container.
Do you have a sample app that does this ? I will try recreate on my side.
Do you have this bug numbers for the issues closed ? I would like to follow
up.
> I let these bugs be closed, because I think that it is not a very
> realistic case, i.e. probably users will not do 1000 times deploy/undeploy
> and don't restart appserver during this period of time. Without "big"
> archives about 2000 such actions (deploy/undeploy/...) can be performed
> without these issues. But also I saw that "big" archives (about 5 MB)
> can kill appserver quickly with 64 bit JVM.
>
>
It seems like this will only be fixed in 3.1.
This worries me quite a bit, since the Eclipse and I think the Netbeans
autodeploy on source change.
This is a potential support nightmare, but that is just my opinion.
regards
Richard.