[
https://issues.apache.org/jira/browse/KAFKA-9450?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17018750#comment-17018750
]
Matthias J. Sax commented on KAFKA-9450:
----------------------------------------
[~vvcephei] Your comment seem to be an orthogonal concern (that I actually
don't share – when we call RocksDB#flush(), it seems to be safe to assume that
RocksDB persisted the data – why do you doubt that RocksDB does not guarantee
this; and if is did not persist it, it would be a RocksDB bug IHMO that should
just get fixed.).
Nevertheless, this ticket is about decoupling of changelog flushing and local
disk flushing – in contrast your comment is about two aspect of local disk
flushing, ie, the data itself and the metadata (ie, checkpoint).
> Decouple inner state flushing from committing with EOS
> ------------------------------------------------------
>
> Key: KAFKA-9450
> URL: https://issues.apache.org/jira/browse/KAFKA-9450
> Project: Kafka
> Issue Type: Improvement
> Components: streams
> Reporter: Sophie Blee-Goldman
> Priority: Major
>
> When EOS is turned on, the commit interval is set quite low (100ms) and all
> the store layers are flushed during a commit. This is necessary for
> forwarding records in the cache to the changelog, but unfortunately also
> forces rocksdb to flush the current memtable before it's full. The result is
> a large number of small writes to disk, losing the benefits of batching, and
> a large number of very small L0 files that are likely to slow compaction.
> Since we have to delete the stores to recreate from scratch anyways during an
> unclean shutdown with EOS, we may as well skip flushing the innermost
> StateStore during a commit and only do so during a graceful shutdown, before
> a rebalance, etc. This is currently blocked on a refactoring of the state
> store layers to allow decoupling the flush of the caching layer from the
> actual state store.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
