We already have plenty of unit tests that have the cognitive burden of shared static state and need for idempotency / defense against order of execution. I think having another tool in our toolbox to profile long-running tests and batch them and/or share a data set makes sense.
On Wed, Sep 15, 2021 at 9:04 AM Berenguer Blasi <berenguerbl...@gmail.com> wrote: > Hi all, > > In CASSANDRA-16951 I have been working on a way to lower the time it > takes to run python dtests in order to reduce CI consumption and cycle > time. > > The approach works by annotating test methods with a @reuse_cluster > annotation. All the test methods following that initial annotation that > use the same annotation will reuse that cluster. Therefore avoiding > starting and stopping the cluster. My initial test show significant > savings. From nice to (1h -> 45m) to great (10m->1.5m). > > The drawback of this approach is that it cannot always be used as it > reduces test isolation. So it needs to be implemented on a per test basis. > > Before I invest more time on the ticket I would be interested in hearing > your opinion on this approach. > > Thx in advance. > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@cassandra.apache.org > For additional commands, e-mail: dev-h...@cassandra.apache.org > >
