Yilin Chen
Yilin Chen
@MyonKeminta I don't think checking the schema when resolving locks would be a final solution even for TiDB. I am looking a concrete plan to avoid it (and it's vital...
I propose making `enable-async-commit` and `enable-1pc` global variables. These two variables are set to 1 in new 5.0 clusters but set to 0 for clusters upgraded from pre-5.0 versions. Upgrading...
The performance improvement is promising. I think we can get started with dealing with the corner cases and compatibility issues.
And I think we can commit your 1PC implementation to the codebase now as long as it's not enabled.
It's a blocker for enabling async apply by default. Maybe not a blocker for async commit GA.
**Update:** With https://github.com/pingcap/tidb/pull/20550, we rely on checks on `max_commit_ts` to guarantee async commit and 1PC transactions conforms the old schema. However, there are still some subtle and not so serious...
Ref: https://twitter.com/the_foreverbell/status/1565361797908930561
Will it work fine with a common handle?
@ekexium `TestWriteConflictPrettyFormat` fails because you add the key to the error.