tiflow
tiflow copied to clipboard
pingcap
drop index did not sync to secondary when simulate secondary failure last for 3mins during add and then drop index on primary
What did you do?
1、restore data for primary and secondary 2、create changefeed and set bdr role for primary and secondary 3、run sysbench on primary and secondary 4、add index and then drop index when add index finished on primary 5、 simulate secondary failure last for 3mins during add and drop index on primary chaos start time：2024-03-06 00:11:33 chaos end time：2024-03-06 00:14:33
ticdc logs: endless-ha-test-bdr-ddl-tps-7230009-1-294.tar.gz
What did you expect to see?
after recover fault, ddl can sync success
What did you see instead?
drop index did not sync to secondary after recover fault
primary：
secondary：
Versions of the cluster
./cdc version Release Version: v8.0.0-alpha Git Commit Hash: fcd4bfa5b89d41b3de663e1966f54bb80a680fe7 Git Branch: heads/refs/tags/v8.0.0-alpha UTC Build Time: 2024-03-04 11:36:45 Go Version: go version go1.21.6 linux/amd64 Failpoint Build: false
current status of DM cluster (execute query-status <task-name> in dmctl)
No response
The issue arises because the CDC executes the add index operation asynchronously to prevent synchronization delay caused by the DDL. As a result, CDC cannot ensure that the previous add index has already entered the downstream DDL pending queue when executing the drop index. In this case, while executing the drop index, the downstream might assume that the index doesn't exist yet. Therefore, CDC disregards this error and assumes that the DDL can be skipped.
