lance6716
> @lance6716 what do you mean by "always mask some SQL mode"? A rough thought, we should remove some SQL mode such as NO_ZERO_IN_DATE when synchronize to downstream, and keep...
> I disagree. I think the behavior of having the DM worker match the upstream SQL mode is 100% correct, but there should be an option to override that if...
maybe the DELETE and other SQL access totally different TiKV regions, after filtering these're less region involved.
/cc @lance6716 @glorv transfer to @GMHDBJD if @glorv is not available
maybe user would depend on this feature wrongly, such as for `t*` in BA list, upstream has `t1~t3`, downstream has `t1~t5`, it's both a little reasonable to remove `t4~t5` or...
> we drop tables based on upstream names. in your example, we should only drop `t1~t3`. > > in other words, we only drop tables we are planning to migrate....
don't know why schema tracker is stuck. could you provide a goroutine stack of DM worker?
DM now prints the original content of file to log. If we modify the content before printing, there's a little chance that the modification will hide an error and make...
Thanks for reporting, currently we can't handle every type of MariaDB. As you said "Data lengh 43940 characters data in that column json, data type BLOB", I don't understand what's...