On Wed, Apr 4, 2018 at 3:55 AM, Gregory Farnum <gfarnum@xxxxxxxxxx> wrote: > Teuthology is not without problems. The next one we discussed was > difficulty even running tests for smaller contributors without access > to an existing lab. This extended a bit to the difficulty of simply > *developing* tests. There isn’t an immediate solution to this problem, > but there are some steps people have taken independently. > > John wrote a “vstart_runner” framework which supports a reasonable > subset of the full teuthology commands and is used throughout the > CephFS test suite (and I assume the manager as well?); this lets you > run tests against a vstart instance with pretty good fidelity to what > happens in the lab. Radoslaw and Orit have a similar thing for RGW > which hasn’t been merged but which that team sometimes uses in test > development: https://github.com/ceph/ceph/pull/20536 . We should unify > these efforts if possible and extend them into more tests whenever > possible. This helps a lot with test development. We’ll of course > never be able to support running a full rados suite on a single > machine (just due to the machine time required), but if we switch > enough tests into a framework that runs in multiple environments > individual contributors can at least run smoke suites on their own, as > well as any new tests they create for their contributions. > PROBLEM TOPIC: unify “tests-without-full-teuthology” frameworks and > use them more extensively going forward. On the ceph-ansible and ceph-volume side of things, we use ceph-ansible itself to deploy test clusters with numerous configurations and run Python unit tests against them. If one of the use cases here is to aid fast(er) testing when developing, then this is certainly a viable way as long as one can point the other tools to the running cluster (e.g. vstart_runner.py ) The test harness doesn't involve any custom and reuses common libraries and tools like: Vagrant, libvirt/virtualbox, and tox. It supports both libvirt and virtualbox, so it is trivial to have a local cluster running to test against. This is also fairly well documented here http://docs.ceph.com/ceph-ansible/master/testing/index.html > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
