Not the answer you need?
Register and ask your own question!

unexpected high memory usage using ndbmtd MySQL Cluster ndbcluster

jl3128jl3128 EntrantCurrent User Role Participant
I am experiencing higher system memory usage than expected when using ndbmtd instead of ndbd. By my count from config.ini each node should be using around 13GB of memory, but looking at the ndbmtd parent process it has consumed around 36GB instead. I know that the RedoBuffer is a "per LDM" value and that each LDM thread is using the value from RedoBuffer, but are there other values that this is true for? It's consuming more than twice what I expect. I've used ndb_config to make sure that the "actual live" values match config.ini values. Perhaps it has something to do with my ThreadConfig and extra threads consuming memory? Please school me!

version: mysql-5.6.24 ndb-7.4.6

# ps aux |grep ndbmtd
root 2031 28.1 47.4 38620704 35290504 ? Sl May06 253:19 /usr/bin/ndbmtd --ndb-nodeid=11

# free
total used free shared buffers cached
Mem: 74377352 38038176 36339176 604 216968 1508032
-/+ buffers/cache: 36313176 38064176
Swap: 3997692 0 3997692

from config.ini

NoOfReplicas = 2
DataMemory = 6G
IndexMemory = 1G
SharedGlobalMemory = 2G
DiskPageBufferMemory = 2G
RedoBuffer = 512M
FragmentLogFileSize = 64M
NoOfFragmentLogFiles = 128
Sign In or Register to comment.

MySQL, InnoDB, MariaDB and MongoDB are trademarks of their respective owners.
Copyright ©2005 - 2020 Percona LLC. All rights reserved.