Closed lppsuixn closed 1 year ago
I have the same problem.It will cause the data to be incorrect.
Thanks @lppsuixn and @jameswangchen for reporting!
Consider paimon sink to add an option so that if the sequence_number provided does not meet the precision, sink goes and makes up a nanos after it.
@JingsongLi I'm willing to task this, could you assign it to me please?
@JingsongLi I'm willing to task this, could you assign it to me please?
Okay, can you think about how to design this API first? How to splice the user's sequence_field and nano_time?
@JingsongLi I'm willing to task this, could you assign it to me please?
Okay, can you think about how to design this API first? How to splice the user's sequence_field and nano_time?
Yes, thanks for reminding~I'll think about it.
@schnappi17 Hello,what is the status of this issue?
@schnappi17 Hello,what is the status of this issue?
Hi @jameswangchen, I'm working on this, will submit PR this week.
Search before asking
Motivation
When consuming CDC data, we use sequence.field to avoid disorder. However, since MySQL's execution time is at the level of seconds, there may be issues with the same ID and same execution time.
Solution
add an auto-incrementing "inner-sequence-number" field in the SystemColumns. When the sequence numbers are the same, can compare the inner-sequence-numbers to determine the order.
Anything else?
No response
Are you willing to submit a PR?