Kyle Kingsbury

Also keep in mind that, like, materializing test results and histories in memory (or even on disk) can be *shockingly* expensive--this is one of those rare places where rolling your...

That's a weird one. I don't know if this is a docker thing, but it looooks like whatever went wrong was some sort of issue with ntpdate: `ntpdate[6305]: no server...

> It works only when ntpdate is run on all the nodes before running lein run test. That’s the workaround we’re using now. That is really strange! I'm not sure...

That's a great idea--want to make a PR against verschlimmbesserung?

Never hit that one before, but pprint does a lot of weird stateful stuff with the stm. It's also real slow so maybe we should replace it with fipp or...

Hi Denisa, I'm moving right now and don't have a lot of bandwidth for OSS stuff, but this souuunds like it's something to do with the way the new queue...

Right... Is there a chance you're trying to upgrade those tests to work with a newer version of Jepsen? The API around queue checking changed a while ago to simplify...

For an example of how drains work now, see https://github.com/jepsen-io/jepsen/blob/master/hazelcast/src/jepsen/hazelcast.clj#L230-L234. Contrast that with https://github.com/jepsen-io/jepsen/blob/master/rabbitmq/src/jepsen/rabbitmq.clj#L171-L180, which has to generate a sequence of synthetic dequeue invocations & completions, and call conj-op! to...

Jsch has been nothing but flaky, and I've been working around these bugs for seven-odd years. There is already a retry built in to jepsen.control--you should see log lines about...

That's a possibility, but that raises the question... is this an error that *should* be retried? Usually we get packet corrupt or session is down; an NPE could indicate any...