to property reflect the upper cache utilization before space is reclaimed. The default ``immutable_object_cache_watermark``
now is ``0.9``. If the capacity reaches 90% the daemon will delete cold cache.
-* Scrubs are more aggressive in trying to find more simultaneous possible PGs within osd_max_scrubs limitation.
- It is possible that increasing osd_scrub_sleep may be necessary to maintain client responsiveness.
-
* OSD: the option ``osd_fast_shutdown_notify_mon`` has been introduced to allow
the OSD to notify the monitor it is shutting down even if ``osd_fast_shutdown``
is enabled. This helps with the monitor logs on larger clusters, that may get