Open nurkiewicz opened 9 years ago
Comment by marcingrzejszczak Friday Dec 19, 2014 at 17:49 GMT
From ivy's perspective we might have a problem:
https://issues.apache.org/jira/browse/IVY-735
Maybe... but this is a little bit insane ;) ... we can run grabbing in a Thread that would have certain time to return a result - and if takes too long we would try to interrupt the thread. Don't know if that would actualy be feasible and also sounds really bad :P
Comment by nurkiewicz Thursday Jan 08, 2015 at 20:31 GMT
Let's make it even more insane. I suggest starting in-memory ZooKeeper first, but move downloading stubs and starting WireMock servers into a background thread. This way we don't slow down application startup and after all, application shouldn't assume all dependencies are available right from the beginning.
Of course there is a problem with error handling. Should we ignore problems while downloading stubs (timeouts, other exceptions) or shut down application in some unexpected moment?
Comment by marcingrzejszczak Sunday Jan 11, 2015 at 16:09 GMT
I see only one issue here - what if the tests execute themselves faster then the stubs are downloaded? Your tests will fail even though they shouldn't.
Comment by nurkiewicz Sunday Jan 11, 2015 at 18:19 GMT
Not necessarily. During tests (test
profile?) this feature can be disabled by default. Moreover during integration tests we can somehow ask application from outside whether stubs are ready. This can be done by calling /collaborators
. Of course there will be a flag (e.g. stubrunner.async-startup
) at user disposal.
Issue by szpak Friday Dec 19, 2014 at 17:34 GMT Originally opened as https://github.com/4finance/stub-runner/issues/23
For example when a server is not accessible without a company VPN. It just hangs for a few minute (possible longer - not tested). In the result the end application startup hangs without any sign what is going on.
A fragment of the long thread dump: