You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
After adapting checkpoint V3, there is no need for keeping the poor old checkpoint model. In the 6.2 release, the config use_checkpoint_v3 would always be true. We can move the code of checkpoint V2 from the codebase.
Generally, there are some steps:
Adapt the tests: removing tests spec for checkpoint V2 and fix failure cases when using V3.
subscription_tracer no more needs to support state transforming.
checkpoint_manager::CheckpointV2FlushObserver can be removed now.
Uploading region level checkpoint to meta_client can be removed now.
A Note: Compatibility
Checkpoint V2 isn't officially released.
If some in-use cluster wants to upgrade to V3, we can always stop the task when updating and do a full backup, start the task again after update succeed. There isn't an intermediate state in updating from V2 to V3 in that condition.
The text was updated successfully, but these errors were encountered:
YuJuncen
changed the title
Tracking Issue for Removing Checkpoint V2 Codebase from TiKV
Log-Backup: Tracking Issue for Removing Checkpoint V2 Codebase from TiKV
Aug 1, 2022
ref #13196
Now, all integration test cases uses checkpoint V3.
Removed `test_inflight_message` because it is invalid in V3.
Signed-off-by: Yu Juncen <yujuncen@pingcap.com>
Co-authored-by: Ti Chi Robot <ti-community-prow-bot@tidb.io>
Development Task
After adapting checkpoint V3, there is no need for keeping the poor old checkpoint model. In the 6.2 release, the config
use_checkpoint_v3
would always betrue
. We can move the code of checkpoint V2 from the codebase.Generally, there are some steps:
subscription_tracer
no more needs to support state transforming.checkpoint_manager::CheckpointV2FlushObserver
can be removed now.meta_client
can be removed now.A Note: Compatibility
Checkpoint V2 isn't officially released.
If some in-use cluster wants to upgrade to V3, we can always stop the task when updating and do a full backup, start the task again after update succeed. There isn't an intermediate state in updating from V2 to V3 in that condition.
The text was updated successfully, but these errors were encountered: