Siyao Meng created HDDS-7743:
--------------------------------
Summary: [Snapshot] Implement snapshot disk usage
Key: HDDS-7743
URL: https://issues.apache.org/jira/browse/HDDS-7743
Project: Apache Ozone
Issue Type: Sub-task
Reporter: Siyao Meng
(Excerpt from the design doc)
We could show two sizes of a snapshot (terms borrowed from btrfs):
a. Referenced size: total size of all the keys in the snapshot. This indicates
how much space it would take on the target cluster when replicated. Basically
this is du on the bucket inside the DB checkpoint. Use the existing counter
(space used) implemented for quota.
b. Exclusive size: the size of keys exclusively referenced by this snapshot.
This indicates how much space (times replication factor) can be released from
the cluster when the snapshot is deleted and garbage collected.
(i) The exclusive size calculation can be done with the help of deletedTable in
the next snapshot (or active filesystem’s if it is the last snapshot). Iterate
over next snapshot’s deletedTable (and delayedDeletedTable), count the number
of keys that exist in the current snapshot, and return the sum of those key
sizes.
(ii) The exclusive size of a snapshot can fluctuate (typically increase) when a
neighboring snapshot is deleted because of the deletedTable merge. Still, for a
read-only snapshot, we can calculate this in a background thread and persist
the size. And whenever a neighboring snapshot is deleted, the persisted value
is invalidated and would have to be calculated again (possibly incrementally).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
