In addition, the buildbot configs are locked to prevent port conflicts. I.e. only one build at a time per build server
On Thu, Jan 1, 2015 at 7:13 PM, Alex O'Ree <spyhunte...@gmail.com> wrote: > It's not a port conflict. I think one of the test creates an entity that > other tests depend on and assume are present. Most of the time it works out > ok, but when junit runs the tests in different orders it doesn't guarantee > that that item is present. Just need to track down the order and figure out > what the root cause is > > On Thu, Jan 1, 2015 at 5:40 PM, Kurt Stam <kurt.s...@gmail.com> wrote: > >> The test open a port. Other builds may choose the same port. Currently it >> picks a random port to reduce the chances. Maybe we can change the interval >> it picks a port number from. Do the logs show port conflict errors? >> >> >> > On Jan 1, 2015, at 16:13, Alex O'Ree (JIRA) <juddi-...@ws.apache.org> >> wrote: >> > >> > Alex O'Ree created JUDDI-904: >> > -------------------------------- >> > >> > Summary: intermittent test failures >> > Key: JUDDI-904 >> > URL: https://issues.apache.org/jira/browse/JUDDI-904 >> > Project: jUDDI >> > Issue Type: Bug >> > Components: uddi-tck >> > Affects Versions: 3.2 >> > Reporter: Alex O'Ree >> > Assignee: Alex O'Ree >> > Fix For: 3.3 >> > >> > >> > The tck test cases ran as part of the maven build occasionally fails. >> The root cause is unknown, but it is most likely due to a logic error >> (failure to clean up some entities) and/or (failure to create required >> entities for the test). When the tests run, the fire order of junit tests >> can vary, thus the cause of this issue >> > >> > >> > >> > -- >> > This message was sent by Atlassian JIRA >> > (v6.3.4#6332) >> > >
