You are viewing an old version of this page. View the current version.
Compare with Current
View Page History
« Previous
Version 6
Next »
For this type of test we have subjected the cluster to an average workload for a few hours, always making use of the applications seen in these pages. Let's see how it behaved based on the flavors of the VMs.
$ kubectl autoscale deployment php-apache --cpu-percent=50 --min=1 --max=20
# We started with 2 replicas, then increased to 4
$ kubectl scale deploy infinite-calls --replicas=2
$ kubectl get hpa --watch
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
php-apache Deployment/php-apache 0%/50% 1 20 1 89m
php-apache Deployment/php-apache 209%/50% 1 20 1 89m
php-apache Deployment/php-apache 209%/50% 1 20 4 89m
php-apache Deployment/php-apache 209%/50% 1 20 5 89m
php-apache Deployment/php-apache 110%/50% 1 20 5 90m
php-apache Deployment/php-apache 110%/50% 1 20 10 90m
php-apache Deployment/php-apache 110%/50% 1 20 11 90m
php-apache Deployment/php-apache 47%/50% 1 20 11 91m
.
.
. # Replicas 2 --> 4
php-apache Deployment/php-apache 70%/50% 1 40 11 142m
php-apache Deployment/php-apache 70%/50% 1 40 16 142m
php-apache Deployment/php-apache 64%/50% 1 40 16 143m
php-apache Deployment/php-apache 64%/50% 1 40 21 143m
php-apache Deployment/php-apache 54%/50% 1 40 21 144m
php-apache Deployment/php-apache 51%/50% 1 40 21 145m
php-apache Deployment/php-apache 50%/50% 1 40 21 147m
In the images below, taken from Grafana, the "jump" in the use of the CPU in the stress namespace is evident, caused by the increase in requests to the web application.
EduranceTest (xLarge)