NUMA Overview Dashboard

For each node, this dashboard shows metrics related to Non-uniform memory access (NUMA).


Users who already have General system metrics service monitored and would like to add NUMA metrics need to remove and re-add linux:metrics on the node:

pmm-admin remove linux:metrics
pmm-admin add linux:metrics

Memory Usage

Remotes over time the total, used, and free memory.

Free Memory Percent

Shows the free memory as the ratio to the total available memory.

NUMA Memory Usage Types

Memory waiting to be written back to disk
Memory used for block device bounce buffers
Files which have been mapped, such as libraries

KernelStack The memory the kernel stack uses. This is not reclaimable.

NUMA Allocation Hits

Memory successfully allocated on this node as intended.

NUMA Allocation Missed

Memory missed is allocated on a node despite the process preferring some different node.

Memory foreign is intended for a node, but actually allocated on some different node.

Anonymous Memory

Anonymous memory that has been used more recently and usually not swapped out.
Anonymous memory that has not been used recently and can be swapped out.

NUMA File (PageCache)

Active(file) Pagecache memory that has been used more recently and usually not reclaimed until needed.

Inactive(file) Pagecache memory that can be reclaimed without huge performance impact.

Shared Memory

Shmem Total used shared memory (shared between several processes, thus including RAM disks, SYS-V-IPC and BSD like SHMEM)

HugePages Statistics

Number of hugepages being allocated by the kernel (Defined with vm.nr_hugepages).
The number of hugepages not being allocated by a process
The number of hugepages in the pool above the value in vm.nr_hugepages. The maximum number of surplus hugepages is controlled by vm.nr_overcommit_hugepages.

Local Processes

Memory allocated on a node while a process was running on it.

Remote Processes

Memory allocated on a node while a process was running on some other node.

Slab Memory

Allocation is a memory management mechanism intended for the efficient memory allocation of kernel objects.
The part of the Slab that might be reclaimed (such as caches).
The part of the Slab that can’t be reclaimed under memory pressure
  • Page updated 2021-01-29


Previous topic

*Compare System Parameters* Dashboard

Next topic

*Prometheus* Dashboard

Contact Us

For free technical help, visit the Percona Community Forum.
To report bugs or submit feature requests, open a JIRA ticket.
For paid support and managed or professional services, contact Percona Sales.