Description
while testing out 4.0-rc0, sarowe noted the he was seeing the smoke tester script fail while sanity checking the solr example.
The crux of the issue seems to be...
- the same directory is reused for testing the example in java6 and then java7
- in some versions of cygwin, SIGINT does not do a clean shutdown of jetty+solr (with jvm close hooks that would do a hard commit)
- when solr is shutdown uncleanly (the java6 run), the tlog is used on startup of the next (java7) run.
- tlog recovery concurrent with rapid updates on startup can be problematic, commits may be ignored.
The underlying questions about dealing with tlog recovery and concurrent updates from external clients have been spun off into SOLR-3888. this issue is focusing on changes that should be considered for the smoke tester.