view-utilization - kubectl plugin to view utilization

---

view-utilization kubectl plugin that shows cluster resource utilization. It is written in BASH/awk and uses kubectl tool to gather information. You can use it to estimate cluster capacity and see at a glance overprovisioned resoures with this simple command kubectl view-utilization.

Installation

krew (kubectl plugin manager)

1. Install krew plugin manager for kubectl.
2. Run kubectl krew install view-utilization.
3. Update plugin with kubectl krew upgrade view-utilization

macOS

On macOS, plugin can be installed via Homebrew:

brew tap etopeter/tap
brew install kubectl-view-utilization

Install with Curl

For Kubernetes 1.12 or newer:

# Get latest tag
VIEW_UTILIZATION_PATH=/usr/local/bin
VIEW_UTILIZATION_TAG=$(curl -s https://api.github.com/repos/etopeter/kubectl-view-utilization/releases/latest | grep  "tag_name"| sed -E 's/.*"([^"]+)".*/\1/')

# Download and unpack plugin
curl -sL "https://github.com/etopeter/kubectl-view-utilization/releases/download/${VIEW_UTILIZATION_TAG}/kubectl-view-utilization-${VIEW_UTILIZATION_TAG}.tar.gz" |tar xzvf - -C $VIEW_UTILIZATION_PATH

# Rename file if you want to use kubectl view-utilization or leave it if you want to invoke it with kubectl view utilization (with space between). Underscore between words allows kubernetes plugin to have hyphen between words.
mv $VIEW_UTILIZATION_PATH/kubectl-view-utilization $VIEW_UTILIZATION_PATH/kubectl-view_utilization

# Change permission to allow execution
chmod +x $VIEW_UTILIZATION_PATH/kubectl-view_utilization

# Check if plugin is detected
kubectl plugin list

Dependencies

While we try to be as minimalistic as possible the only dependency is Awk. While GNU Awk (gawk) is recommended, mawk is also supported.

• kubectl
• bash
• awk (gawk, mawk)

Usage

This plugin should be invoked with kubectl command, and will appear as subcommand. It will use the existing context configured in $KUBECONFIG file. You can override context with --context parameter.

kubectl view-utilization
Resource     Requests  %Requests        Limits  %Limits   Allocatable   Schedulable         Free
CPU             43475         81         70731      132         53200          9725            0
Memory    94371840000         42  147184418816       66  222828834816  128456994816  75644416000

Column	Short	Description
Requests	Req	Calculated total pod requests across all namespaces
%Requests	%R	Percentage of total requests against allocatable requests
Limits	Lim	Calculated total pod limits across all namespaces
%Limits	%L	Percentage of total limits against allocatable limits
Allocatable	Alloc	Available allocatable resources
Schedulable	Sched	Resources that can be used to schedule pods; Available for pod requests (allocatable - requests)
Free	Free	Resources that are outside all requests or limits

Example usage:

Human readable format -h

kubectl view-utilization -h
Resource  Req   %R   Lim    %L  Alloc  Sched  Free
CPU        43  71%    71  117%     60     17     0
Memory    88G  37%  138G   58%   237G   149G   99G

Check utilization for specific namespace -n

kubectl view-utilization -h -n kube-system
Resource   Req  %R   Lim  %L  Alloc  Sched  Free
CPU        3.7  6%   4.3  7%     60     57    56
Memory    5.4G  2%  7.9G  3%   237G   232G  229G

Check utilization for node groups using label filters. Example filter results only for nodes in availability zone us-west-2b failure-domain.beta.kubernetes.io/zone=us-west-2b:

kubectl view-utilization -l failure-domain.beta.kubernetes.io/zone=us-west-2b -h
Resource  Req   %R  Lim    %L  Alloc  Sched  Free
CPU        14  64%   24  106%     22      8     0
Memory    30G  33%  47G   52%    89G    59G   42G

Breakdown of node utilization kubectl view-utilization nodes

CPU   : ▆▆▆▆▅▅▇▄▄▆▂▆
Memory: ▇▅▆▇▇▅▅▄▆▇▁▆
                                           CPU                 Memory               
Node                                       Req   %R  Lim   %L  Req   %R  Lim   %L
ip-10-0-0-175.us-east-1.compute.internal   8.1  53%   13   86%  24G  83%  31G  105%
ip-10-0-0-55.us-east-1.compute.internal    6.6  43%   13   90%  19G  64%  22G   76%
ip-10-0-18-238.us-east-1.compute.internal    7  46%   12   85%  24G  82%  25G   86%
ip-10-0-19-235.us-east-1.compute.internal   10  67%   14   93%  27G  92%  29G   98%
ip-10-0-21-0.us-east-1.compute.internal    9.5  63%   12   83%  25G  86%  30G  101%
ip-10-0-28-44.us-east-1.compute.internal   6.9  45%   10   70%  20G  70%  24G   81%
ip-10-0-3-133.us-east-1.compute.internal     6  40%   14   97%  20G  67%  24G   83%
ip-10-0-3-24.us-east-1.compute.internal    5.9  39%   10   66%  17G  57%  19G   63%
ip-10-0-35-119.us-east-1.compute.internal  7.7  51%   10   66%  23G  78%  28G   94%
ip-10-0-39-146.us-east-1.compute.internal   10  66%   13   90%  25G  84%  30G  101%
ip-10-0-40-184.us-east-1.compute.internal  3.6  23%  5.7   37%  11G  17%  13G   21%
ip-10-0-42-24.us-east-1.compute.internal   6.6  43%   13   90%  22G  76%  26G   88%

Overview of namespace utilization kubectl view-utilization namespaces

kubectl view-utilization namespaces -h
             CPU        Memory
Namespace     Req  Lim   Req   Lim
analitics     6.6   10   14G   21G
kube-system   3.5  4.2  5.1G  7.6G
lt             13   21   27G   42G
monitoring   0.35  3.5  1.8G  3.5G
qa             13   21   27G   42G
rc            6.6   10   14G   21G

Output to JSON format.

kubectl view-utilization -o json | jq
{
  "CPU": {
    "requested": 43740,
    "limits": 71281,
    "allocatable": 60800,
    "schedulable": 17060,
    "free": 0
  },
  "Memory": {
    "requested": 94942265344,
    "limits": 148056834048,
    "allocatable": 254661525504,
    "schedulable": 159719260160,
    "free": 106604691456
  }
}

---

Simplify workflow with aliases

Add to your ~/.bash_profile or ~/.zshrc

alias kvu="kubectl view-utilization -h"

Now you can use kvu alias to quickly show resource usage

Example commands:

kvu
kvu namespaces
kvu -n kube-system

---

Change log

See the CHANGELOG file for details.

Developing

1. Clone this repo with git
2. Test locally with kubectl pointing to your cluster (minikube or full cluster)
3. Run unit tests make test