Description:
Memory usage statistics in ndbinfo.memoryusage table are way off
I am running a single node setup of (mysql-cluster-gpl-7.5.12-linux-glibc2.12-x86_64) for testing with following configuration
DataMemory=200M
IndexMemory=200M
mysql> select * from ndbinfo.memoryusage;
+---------+---------------------+--------+------------+-----------+-------------+
| node_id | memory_type | used | used_pages | total | total_pages |
+---------+---------------------+--------+------------+-----------+-------------+
| 1 | Data memory | 786432 | 24 | 315621376 | 9632 |
| 1 | Index memory | 278528 | 34 | 327680 | 40 |
| 1 | Long message buffer | 655360 | 2560 | 67108864 | 262144 |
+---------+---------------------+--------+------------+-----------+-------------+
Both the data memory and index memory are way off.
However, when I run an older version (mysql-cluster-gpl-7.5.5-linux-glibc2.5-x86_64) I get the correct results.
mysql> select * from ndbinfo.memoryusage;
+---------+---------------------+--------+------------+-----------+-------------+
| node_id | memory_type | used | used_pages | total | total_pages |
+---------+---------------------+--------+------------+-----------+-------------+
| 1 | Data memory | 589824 | 18 | 209715200 | 6400 |
| 1 | Index memory | 278528 | 34 | 105906176 | 12928 |
| 1 | Long message buffer | 655360 | 2560 | 67108864 | 262144 |
+---------+---------------------+--------+------------+-----------+-------------+
This bug is causing serious problems for us as our application monitors ndbinfo.memoryusage
How to repeat:
as above
