Container resource usage

From ISPWiki
Jump to: navigation, search

Container resource usage statistics collects the following data :
parameters that are used both for VM and cluster nodes:

  • incoming/outgoing bandwidth, GiB
  • CPU, %
  • system load average

container specific parameters:

  • RAM allocation, %
  • RAM allocation problems
  • CPU allocation problems
  • Open files
  • Open processes
  • IOPS

Statistical information is collected once in 5 minutes. It can be collected per 5 minutes, 1 hour and 1 day. You can select a desired period to view resource consumption by your container.

Information on RAM, files/processes

The information is taken from the file /proc/bc/CTID/resources.

CTID - container's identifier.
The amount of RAM consumed is defined by beancounter of privvmpages. The held filed shows actual consumption of RAM, the limit field - limit for VM RAM allocation, the failcnt field represents RAM allocation problems. Percentage of RAM consumption is calculated as follows held/limit*100
numfile beancounter is used for getting information on files. held - the number of currently open files, failcnt - allocation problems.
numproc beancounter is used for getting information on processes. held - the number of currently running processes, failcnt - allocation problems.

Information on I/O operations

The information is taken from " /proc/bc/CTID/iostat.
CTID - container's identifier.
The requests completed field is the number of operations performed.