1

My cluster has one MASTER AGGREGATOR and one LEAF. After running two months, the CPU usage in LEAF is very high, almost at 100%. So, is this normal?

By the way, its size is 545 MB for table data.

LEAF status figure

Peter David Carter
  • 2,548
  • 8
  • 25
  • 44
downkang
  • 21
  • 3

1 Answers1

1

This is not normal for MemSQL operation. Note that the Ops console is showing you all CPU use on that host, not just what MemSQL is using. I recommend running 'top' or similar to determine what process(es) are consuming resources. You can also run 'SHOW PROCESSLIST' on any node to see if there is a long-running MemSQL process.

Douglas
  • 21
  • 2
  • Thank you for your reply.I runned 'top' and here is the list: 23060(PID) memsql(User) 99%(CPU) 48%(Memory) leaf-3306/memsqld(Command). – downkang Feb 19 '16 at 00:38
  • OK. Looks like you are running MemSQL 'cluster-in-a-box', that is master agg and leaf on one logical host. Cloud or physical hardware? Recommend upgrade if you're using an older version. Check running threads with [SHOW PROCESSLIST](http://docs.memsql.com/latest/ref/SHOW_PROCESSLIST/). Restart the cluster `memsql-ops memsql-restart -all`. Memory stable or creeps back up? – Douglas Feb 19 '16 at 05:48
  • Master agg and leaf on two different virtual host but on one physical hardware.MemSQL version is 4.1.2.I runned 'show processlist', state is idle for 400 running threads.My another question is http://stackoverflow.com/questions/35055552/memsqlwhy-memory-is-so-high. The two problems are on the same running environment . – downkang Feb 20 '16 at 08:58
  • What happens when you restart the cluster? How many connections are you opening? Can you share details on your workload? – Douglas Feb 26 '16 at 01:47
  • It is OK when i restart the cluster.Thank you very much! – downkang Mar 03 '16 at 02:54