A final update on this.
It was a configuration error in my part.
sorry for the noise.
On 11/25/21 16:27, Vassilis Virvilis wrote:
Hi again,
I think confirmed that is the same issue. I have updated the bug
https://jira.mariadb.org/browse/MDEV-26552
Now the bug has two reproducer scripts
Hi again,
I think confirmed that is the same issue. I have updated the bug
https://jira.mariadb.org/browse/MDEV-26552
Now the bug has two reproducer scripts attached
a) for the index creation
b) for the faulty group by in filesort mode
My hunch says that is the same bug because in filesort
Hi Sergei,
Thanks for the detailed and insightful answer.
I get it now. So I could invoke the index by FORCE INDEX. Interesting...
MDEV-8306 looks interesting!
The real problem I had is that I get correct results when index is utilized and
wrong with filesort.
My table is more than 500M
Hi, Vassilis!
On Oct 12, Vassilis Virvilis wrote:
>
> I managed to create a trivial reproducer with a 10 rows table.
>
> If I don't specify LIMIT the plan goes to filesort.
> If I specify LIMIT <= 9 the plan goes to utilize the index
> If I specify LIMIT >= 10 (table rows) the plan foes to
Hi,
I am new here,
Not sure if this is the appropriate place. If not, any pointers are much
appreciated.
Not sure if this is a bug that I should report it. Any input on that matter
will be much appreciated also.
Platform: Debian unstable/SID, MariaDB: 10.5.12-MariaDB-1 SSD disks joined in a
5 matches
Mail list logo
