Steven Niemitz

sorry I was OOO last week, I'll take a look at this today!

> No, worries. Hopefully you had a good vacation. I did, thank you! this looks good to me, always great to get rid of static state as well.

> The other part of the change makes sense to reduce byte[] copies by using ByteString. > > CC: @tudorm Maybe I'll pull the ByteString refactoring stuff out into another...

I opened PR #17802 with just the ByteString changes if you want to get those first.

> Have you tried the pipeline visitor using a SnappyCoder that wraps existing encodings? now that we have compression working well end-to-end in our internal fork (and a good baseline...

If you could attach your logs from mesos/kafka-mesos/etc that'd be helpful in debugging this. Its possible the vagrant script is just broken though.

What error did you get when starting up? Would going to 1.0.0 be enough to fix it? Otherwise, what's the minimum master version the 1.3.0 driver can support?

I think this is fine at this point. I'll add a note that the driver version can be changed if needed. This breaks a lot of the unit tests though,...

Do you see anything interesting in the mesos agent logs? (slave-6 in this case). It looks like its exactly 10 seconds after the ACCEPT was sent, which seems suspiciously like...

That's odd that you weren't getting those verbose logs before. Anyways I suspect the issue is the executor crashing on launch ``` 2017-03-14 15:06:28,299 INFO KafkaMesosScheduler] [statusUpdate] kafka-0-1505ef66-a712-404f-95ed-2f97ee60edd3 TASK_FAILED slave:#22-S3...