When has_relevant_range_on_this_shard() found a relevant range, it will unnecessarily iterate through the end. Verified manually that this could be thousands of pointless iterations when streaming data to a node just added. The relevant code could be simplified by de-futurizing it but I think it remains so to allow task scheduler to preempt it if necessary. Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com> Message-Id: <20200220224048.28804-2-raphaelsc@scylladb.com>
14 KiB
14 KiB