Control groups contain several parts that together make it possible to set CPU quotas for processes Freezer and CPU controller. freezer To stop resume the process there is Freezer. This subsystem was introduce because sending SIGSTOP SIGCONT signals is an unreliable way to stop and resume a process proof . The problem with these signals is that they are visible to the process they are sent to. Moreover the SIGCONT signal can be intercepte and handle custom. So stop resume can break an application that intercepts signals.
Both The Enterprise Segment And Startups
The Freezer subsystem in turn uses internal kernel mechanisms to stop and resume processes. Therefore its work is not visible to the process on which it affects. CPU controller To manage and monitor CPU usage cgroups have a CPU controller . It allows you to set scheuler options for a group of processes. Alternatively two parameters can be use to Bermuda Email List limit processes by CPU consumption cpu fs period us and cpu cfs quota us. The first parameter specifies the period of time after which the reallocation of resources should occur timeslice. The second parameter cpu cfs quota us sets the time available for work within this period. All times are specifie in microseconds the us suffix hints at this but its better to clarify . You can also use the cpu shares parameter. This file contains for each process group its weight which determines how much processor time the group can use.
The Private Already Has Its Own Specifics
By default this parameter is units for each group. The most interesting thing is that the real effective weight is determine base on the sum of all weights at the current level of the hierarchy wrappe up normally . You can more clearly show the features IT Cell Number of cpu shares using the following illustration Here the behavior is that of the host resources will be available to the left virtual machine and only will be available to the right one. In this case the available CPU resource for the right virtual machine will be divide equally between each of its cores. The actual value of cpu shares of the cores of the right machine will be or . of the total time. The effective cpu shares value of the left machines kernel will be or of the total time.