I had a look at it. It is caused by a deadlock when a temporary dataset triggers a flush. It wasn’t detected when I introduced a change that eliminates FLUSH logs generation for temporary indexes because recovery test cases were broken.
I submitted a fix for it [1] and added a runtime test case. The build should succeed after it gets merged. Cheers, Murtadha [1] https://asterix-gerrit.ics.uci.edu/#/c/785/ On Mon, Apr 4, 2016 at 9:18 AM, Till Westmann <[email protected]> wrote: > Hi, > > since the last successful test run that took 55 mins [1] we seem to have > more than doubled the runtime to time out at 2 hours [2]. > Does anybody have an idea what caused this? > > Cheers, > Till > > [1] https://asterix-jenkins.ics.uci.edu/job/asterix-integration-tests/124/ > [2] https://asterix-jenkins.ics.uci.edu/job/asterix-integration-tests/130/ >
