I don't see this in the test runs, and the only box I have reproduced a
failure on is due to an inability of the client to reuse the 4446 port,
and so the client remoting layer fails to startup its streaming port.
Its not even getting to the point of trying a deployment.
I'm now looking into why the current runs are failing to shutdown:
tests-report:
[tar] Building tar:
/Users/svn/JBossHead/jboss-head/testsuite/output/2008-08-31.21-39.tgz
BUILD FAILED
/Users/svn/JBossHead/jboss-head/testsuite/build.xml:969: Some servers
failed to shutdown cleanly.
Total time: 179 minutes 36 seconds
[589][valkyrie: testsuite]$
I have reproduced this, but the server is in fact shutting down. The
last servers run are the cluster-profilesvc-0/1 configs, and their logs
show clean shutdowns. I guess we need more debug output from the server
task as to what it thinks is not shutting down.
Scott Stark wrote:
It should not matter because the scanner asks the profile for the
new/changed deployments. It no longer looks directly to the deploy
directory. On OSX the profile service tests are all passing. I'm
updating a fedora box with the current workspace to see its behavior.
Dimitris Andreadis wrote:
> Shouldn't we use a different upload dir in data/ or tmp/ to avoid
> HDScanner seeing those deployments?
>
> Scott Stark wrote:
>> They are uploaded. They should be deleted though on undeployment.
_______________________________________________
jboss-development mailing list
jboss-development(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/jboss-development