bq: batch size != queue size
I wasn't confusing the two, they just happen to be the same numbers. What I'm after here is something on record as to whether we expect the anecdotal evidence of faster indexing by increasing batch size to be affected by streaming, making changing the batch size question less relevant.
I suppose with Joel's/Mark's changes to CloudSolrServer, much of the question goes away anyway, and this JIRA will then primarily affect the leader forwarding the updates to the followers.
Of course all bets are off if a user directly uses SUSS. Seems like a "best practice" is use CloudSolrServer and if you really insist on NOT using it, then this JIRA will keep you from deadlock.