Correct the implemention of drainer mysql checkpoint

Feature Request

Is your feature request related to a problem? Please describe:

There are some problems with implemention of drainer mysql checkpoint, one mysql checkpoint example:

mysql> select * from tidb_binlog.checkpoint;
+---------------------+---------------------------------------------------------------------------------------------------------+
| clusterID           | checkPoint                                                                                              |
+---------------------+---------------------------------------------------------------------------------------------------------+
| 6782769820199954307 | {"commitTS":413989456971300865,"ts-map":{"master-ts":413989358457585669,"slave-ts":413989440066945029}} |
| 6782784420681659503 | {"commitTS":414399178233348097,"ts-map":{"master-ts":414398077844324967,"slave-ts":414398082560032770}} |
+---------------------+---------------------------------------------------------------------------------------------------------+
2 rows in set (0.00 sec

there're two columns in checkpoint table

• clusterID which is primary key
• checkpoint which contain the specific checkpoint value
  , and clusterID is fetched from PD service of theupstream tidb cluster

the implementation brings three problems:

• how to fetch clusterID while the upstream cluster is completely down
• if user want to deplay multiple drainer to replicate different tables from one upstream tidb cluster to the same downstream, they must specify different checkpoint schema and table. It's unreasonable, and user can't always figure out the weird logic

Describe the feature you'd like:

maybe we let user to specify the node-id for every drainer, and store node-id into checkpoint

The issue names no files or tests. Start by locating drainer's MySQL checkpoint handling and its PD clusterID lookup, then examine how upstream unavailability and multiple drainers sharing one downstream are handled. Done should be an agreed checkpoint identity and storage design that addresses both cases, with tests covering them.