Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.5.8
-
None
-
primary Production env 5x Rhel 8 24G ram 8 cpus galera cluster
primary dev 3x rhel 8 8G 8 CPU galera cluster
Both of the above are vms
developer dev single node Cent OS 7
Description
In the past week we have been having issues with some statements using inner joins, in one case a join will sit there pegging the CPU of the node it is on to 100% until stopped, we typically stop the process after 30m or so. The query for this is
SELECT COUNT(*) FROM `resources` INNER JOIN `is_part_ofs_resources` ON `resources`.`id` = `is_part_ofs_resources`.`resource_id` WHERE `is_part_ofs_resources`.`is_part_of_id` = 39 AND `resources`.`discarded_at` IS NULL |
This should have about 13k rows returned . The query has been tested to work fine in the latest versions of 10.1, 10.2, 10.3 and 10.4, and only hangs in 10.5
If we change the query to id = 1 it runs fine with only about 1k returned.
Attached is the json output of the query analizer for the smaller version of is_part_of_id=1