Details
Description
Ignite 2.9 introduces a new feature to take consistent cluster snapshot.
However, there is currently no way for the user to restore a separate cache group from such a snapshot.
As a first step, we must provide the ability (CLI command and public API) to restore the separate cache group(s) without considering the current partition distribution and rely on rebalancing to achieve the ideal distribution.
This process consists of the following steps:
- Make sure that all partitions of the cache group are available in the cluster and that there are no conflicts in the saved cache configurations.
- Make sure the target cache group does not exist (user must manually destroy the cache before restoring)..
- Move the cache data files locally on each node with the cache data snapshots.
- Merge the binary metadata from the snapshot on each node.
- Dynamically start the restored cache group.
Attachments
Issue Links
- causes
-
IGNITE-15170 Automatic snapshot restore operation fails if node consistent ID is not defined
- Resolved
- is related to
-
IGNITE-14070 Implement API for custom snapshot lifecycle handlers (extensions).
- Resolved
-
IGNITE-14723 Add CLI command to restore a cache group from the snapshot.
- Resolved
-
IGNITE-14794 Add JMX command and metrics for automatic snapshot restore operation.
- Resolved
-
IGNITE-14933 Implement metrics for a snapshot restore operation
- Resolved
-
IGNITE-14722 Document the ability to automatically restore a cache group from the snapshot.
- Resolved
- requires
-
IGNITE-14352 Add the ability to check the consistency of individual cache groups in a snapshot.
- Resolved
- links to