Lchangliang opened a new pull request, #22557:
URL: https://github.com/apache/doris/pull/22557

   ## Proposed changes
   
   Issue Number: close #xxx
   
   <!--Describe your changes.-->
   
   Associated with https://github.com/apache/doris/pull/18822, BE shouldn't see 
the prefix "_shadow_" in column name. It still has some bug when doing schema 
change.
   During altering column from int to double, insert more values will doubly 
write to new/old tablets and it will coredump because the insert value doesn't 
match with new tablet column.
   ```
   For example:
   table schema: c1 int, c2 int
   alter c2 from int to double
   insert values (1, 1)
   In BE, the tuple slots: c1 int, c2 int, _shadow_c2 double
   new tablet schema: c1 int, c2 double.
   ```
   Before this fix, we only use column name to match so that new tablet will 
read the "c2 int" slot. And then the int value doesn't match double column.
   Fix it by using <column_name, type> to match slot.
   ```
   Now for example:
   table schema: c1 int, c2 int
   alter c2 from int to double
   insert values (1, 1)
   In BE, the tuple slots: c1 int, c2 int, c2 double (remove prefix "_shadow_" 
too)
   new tablet schema: c1 int, c2 double.
   ```
   it will use (c2, double) to find the correct slot.
   
   ## Further comments
   
   If this is a relatively large or complex change, kick off the discussion at 
[[email protected]](mailto:[email protected]) by explaining why you 
chose the solution you did and what alternatives you considered, etc...
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to