Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.2.13
-
None
-
CentOS 7 (3.10.0-693.17.1.el7.x86_64) on Virtual Box VM with 8 GB of RAM and 16 GB of storage. Note, I am also seeing the issue with a SuperMicro server running CentOS 7 (3.10.0-693.17.1.el7.x86_64) bare metal.
Description
I am attempting to move to the MariaDB 10.2 line for my production application. Previously my upgrade to 10.2 was blocked by MDEV-14857. Now in 10.2.13 this issue has been resolved, however I am running into a new problem. I have a nightly job that runs in a stored procedure that fails in MariaDB 10.2.13. The error I am getting is below:
ERROR 1114 (HY000): The table '#sql-daf_13' is fulldone
I realize that this error is normally due to a configuration issue, however look at the end of the error string, it says "done" after full.
I have narrowed down the issue using the following reproduction steps:
- I have uploaded a sql script to ftp.askmonty.org/private called brycejlowe-permtest.sql
- Install MariaDB 10.2.13 and source in the file, it creates a schema called scratch and a table called permtest1 which has about 422,000 rows in it.
- Execute the following statements against that table:
CREATE TEMPORARY TABLE scratch.temptest1 LIKE scratch.permtest1;
INSERT INTO scratch.temptest1 SELECT * FROM scratch.permtest1;
Â
CREATE TEMPORARY TABLE scratch.temptest2 SELECT * FROM scratch.temptest1;
ALTER TABLE scratch.temptest2 ADD INDEX `part`(`Part_no`);
The result in MariaDB 10.2.8 through 10.2.12 is a temporary table scratch.temptest2 with an index called part. However the result in 10.2.13 is the error:
ERROR 1114 (HY000): The table '#sql-daf_13' is fulldone