chunks that are not persisted yet can be cleared about metrictank HOT 2 CLOSED

This is quite relevant to another problem i have been thinking about.
When we add multi-node support, we need a way to persist older chunks that have not already been persisted before persisting newer chunks.

I think the answer to both these problems is to:

• when persisting a chunk, first check that the previous chunk has been persisted. This logic will then cascade through all chunks ensuring that all chunks have been persisted.

from metrictank.

I believe this no longer is a problem: we now call persist() before clearing a chunk, and even when we clear a chunk (which doesn't really clean anything up) and reuse the slot for a new chunk, the CWR keeps a reference to the old chunk.

from metrictank.