mirror of
https://github.com/ceph/ceph
synced 2025-01-11 05:29:51 +00:00
b4ca5ae462
The newly introduced 'device-class' can be used to separate different type of devices into different pools, e.g, hdd-pool for backup data and all-flash-pool for DB applications. However, if any osd of the cluster is currently running out of space (exceeding the predefined 'full' threshold), Ceph will mark the whole cluster as full and prevent writes to all pools, which turns out to be very wrong. This patch instead makes the space 'full' control at pool granularity, which exactly leverages the pool quota logic but shall solve the above problem. Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn> |
||
---|---|---|
.. | ||
all | ||
% | ||
rados.yaml |