Support check committed txns before catalog drop meta, like db, table, partition etc #4029
Conversation
|
for #3988 |
|
If I use |
@morningman in fe, the txn will be publish timeout, and publish version daemon will try to finish transaction, it find that table have been dropped and will remove table committed info from txn, the txn will be visible. In be, it seems that txns wiil be cleaned by reporter handler? |
caiconghui
changed the title
|
I found if perform DROP database force, it's committed version will be recovered by RECOVER command? |
@chaoyli I will fix it, make force drop operation do not recycle meta. |
fe/src/main/cup/sql_parser.cup
Outdated
| RESULT = new DropPartitionClause(ifExists, partitionName, isTempPartition); | ||
| RESULT = new DropPartitionClause(ifExists, partitionName, isTempPartition, !isTempPartition); | ||
| :} | ||
| | KW_DROPP opt_tmp:isTempPartition KW_PARTITION opt_if_exists:ifExists ident:partitionName |
There was a problem hiding this comment.
DROPP is a not a good syntax. It's better to use
DROP TABLE $table_name FORCE
There was a problem hiding this comment.
@chaoyli I has fixed it before. old commit version?
3 participants
This PR is to ensure that dropped db , table or partition can be with normal state after recovered by user. Commited txns can not be aborted, because the partitions's commited versions have been changed, and some tablets may already have new visible versions. If user just don't want the meta(db, table or partition) anymore, just use dropp instead of drop to skip committed txn check.