...
| Code Block | ||||||
|---|---|---|---|---|---|---|
| ||||||
$ kubectl autoscale deployment php-apache --cpu-percent=50 --min=1 --max=50 $ kubectl scale deploy infinite-calls --replicas=15 $ kubectl get deploy NAME READY UP-TO-DATE AVAILABLE AGE infinite-calls 15/15 15 15 5h4m php-apache 22/42 42 22 5h6m $ kubectl get hpa --watch NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE php-apache Deployment/php-apache 0%/50% 1 50 1 49m php-apache Deployment/php-apache 250%/50% 1 50 1 50m php-apache Deployment/php-apache 250%/50% 1 50 4 50m php-apache Deployment/php-apache 250%/50% 1 50 5 51m php-apache Deployment/php-apache 212%/50% 1 50 5 51m php-apache Deployment/php-apache 212%/50% 1 50 9 51m php-apache Deployment/php-apache 197%/50% 1 50 9 52m php-apache Deployment/php-apache 197%/50% 1 50 18 52m php-apache Deployment/php-apache 197%/50% 1 50 36 52m php-apache Deployment/php-apache 100%/50% 1 50 36 53m php-apache Deployment/php-apache 100%/50% 1 50 42 53m php-apache Deployment/php-apache 99%/50% 1 50 42 54m php-apache Deployment/php-apache 102%/50% 1 50 42 55m php-apache Deployment/php-apache 102%/50% 1 50 42 56m php-apache Deployment/php-apache 101%/50% 1 50 42 57m php-apache Deployment/php-apache 100%/50% 1 50 42 58m php-apache Deployment/php-apache 99%/50% 1 50 42 59m php-apache Deployment/php-apache 101%/50% 1 50 42 60m php-apache Deployment/php-apache 100%/50% 1 50 42 61m php-apache Deployment/php-apache 98%/50% 1 50 42 63m php-apache Deployment/php-apache 100%/50% 1 50 42 64m php-apache Deployment/php-apache 101%/50% 1 50 42 65m php-apache Deployment/php-apache 97%/50% 1 50 42 66m php-apache Deployment/php-apache 101%/50% 1 50 42 67m php-apache Deployment/php-apache 100%/50% 1 50 42 68m php-apache Deployment/php-apache 0%/50% 1 50 42 69m php-apache Deployment/php-apache 0%/50% 1 50 42 74m php-apache Deployment/php-apache 0%/50% 1 50 1 74m |
We can deduce from these graphs, observing the ordinate axis, the percentage of CPU used. The maximum CPU available is given by the sum of the CPUs of the worker nodes alone, which bear all the workload (it is preferable not to perform operations on the master to protect it from any crashes). We have respectively 24, 12 and 6 CPUs available based on flavor and, on average, a percentage equal to approximately 115%, 79% and 82% of use.
...