Since most users will rely upon the async flush of old segments, perhaps we
can leave the default as it is and just make it clear in the documentation?
Thanks,
Jun
On Wed, Mar 5, 2014 at 10:12 AM, Jay Kreps <jay.kr...@gmail.com> wrote:
> It looks like we disable the flush check by default
> /* the frequency in ms that the log flusher checks whether any log needs to
> be flushed to disk */
> val logFlushSchedulerIntervalMs = props.getLong("
> log.flush.scheduler.interval.ms", Long.MaxValue)
>
> This means that if you set a flush time for a log
> bin/kafka-topics.sh --zookeeper localhost:2181 --alter --topic my-topic
> --config flush.ms=30*1000
>
> It actually won't run because we never schedule the check. This is fixable
> if you discover this magical property...
>
> Should we fix this for 0.8.1 or just punt until 0.8.2?
>
> -Jay
>
