High CPU on one node

Hello!

I have a problem with one node within cluster.
htop shows that vertica process gone mad and utilizes all CPU cores at near 100%
LA keep performance above number of CPU cores.
All other nodes at normal (10 times smaller) rates of LA and CPU utilization.
How to find out whats going on with vertica process on bad node.
I cant see anything suspicious in vertica.log on that node, or queries with high resource acquisitions.

Can you help me to find out what vertica process really doing that causes such CPU performance? Is there any best practices or recomendations in such cases

Comments

  • Use system_services table to see if there is specific service that is running abnormally on a node in question. I would specifically check for TM Mergeout and analyzeRowCount service .

  • Thanks. system_services and system_sessions realy helped. 400 system sessions stalled that node. Turned out it was hardware issue. Memory chip gone wild causing vertica to go mad too

Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Emoji
Image
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file