uschindler commented on pull request #1842: URL: https://github.com/apache/lucene-solr/pull/1842#issuecomment-689410718
> I don't want to block the changes here, but just think we should make a followup to implement the same test, but with clients in the same JVM. > > There are two separate cases for NIO locks (held by same VM vs held by another process), so we should stress test them both. > Past experience has shown that whatever lock testing we have, it is somehow never enough. I think this could be a separate issue. Basically, this just "restores" the test in master, which was done in Ant only. There are no functional changes. But this new code is easy to adapt, there are several possibilities: - We can clone the whole test and just replaces `List<Process> processes` by a thread pool and instead of spawning a JVM we just call main() in a thread. Cleanup is more or less the same. - We can modify the current test to not only spawn multiple JVMs, but aditionally also add a thread. We can even randomize the number of clients/threads (currently fixed to 2 clients, but it's just a change in test config). In both cases, the policy also needs the "connect" permission. An alternative would be to allow client/server to alternatively communicate through a PipedInput/OutputStream (if its in same JVM). Some questions before I commit this: - Should I rename the test to TestStressLockFactories? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
