Thanks all, I'll know enough to look next time...... Erick
On Tue, Mar 5, 2013 at 8:08 AM, Dawid Weiss <[email protected]>wrote: > > >> Just to make it worse (better?) since Uwe pointed me at the need to close >> streams that weren't XML configs, I haven't seen this particular test >> failure. >> One additional point of clarification is that by "hard coding" the >> output, I >> really meant logging it without using the verbose() construct. >> > > To add to the confusion, tests.verbose system property is not really > related to how the test framework works, Erick. It is still *used* to > initialize certain flags in the code and these flags will dump more output > if enabled -- this is a remnant of the past when all output was just pumped > to the console and this was a way to limit the noise (my guess). So if you > enable this property you'll get more output but it doesn't affect what we > were talking about (output directed to the console or not). > > >> So if I'm reading this right, I should be able to go to the server where >> the >> test failed and see my output, if I get to it before it's been wiped >> since this >> test is a _failed_ test. >> > > If so then all of its output should be dumped to the console upon failure > and should be available on jenkins in the execution log. Much like for this > test, for example: > > > https://builds.apache.org/job/Lucene-Solr-Tests-trunk-java7/3783/consoleText > (grep for org.apache.solr.cloud.BasicDistributedZkTest and you'll see all > of its stderr/stdout piped right underneath). > > Like Uwe said -- those plain text files I mentioned contain all tests and > all output from all tests regardless if they succeeded or not; I don't > think it's really worth storing them if a test passed (perhaps there is a > way to store the latest X builds though if Uwe has a lot of space on that > beast machine of his). > > Dawid > >
