Repository files navigation Build:
./build.sh build
Publish:
./build.sh publish
Deploy:
helm upgrade -i gluster ./app-gluster
After all gluster pods (3 by default) are up and running, connect them into a cluster:
kubectl exec gluster-app-gluster-0 -- gluster peer probe gluster-app-gluster-1.gluster-app-gluster.default.svc.cluster.local
Unfortunately there is an issue with reverse DNS lookup for StatefulSet pods in Kubernetes, so after peer discover
gluster peer hostnames must be fixed in some peers at least.
Fortunately this is a one-time procedure:
kubectl exec gluster-app-gluster-0 -- /bin/bash -c 'sed -r -n -i -e '\''/^\s*hostname[0-9]+\s*=.*/ !p'\'' -e '\''/^\s*hostname[0-9]+\s*=gluster-app-gluster-[0-9]+\.glust er-app-gluster\.default\.svc\.cluster\.local$/ p'\'' /var/lib/glusterd/peers/*; systemctl restart glusterd'Now you can verify that all peer addresses are correct:
kubectl exec gluster-app-gluster-0 -- gluster pool list
Create a gluster volume:
kubectl exec gluster-app-gluster-0 -- mkdir -p /bricks/vol1
... and check its status:
kubectl exec gluster-app-gluster-0 -- gluster volume info
Now we can create corresponding Gluster Kubernetes persistent volume, ...
kubectl apply -f - <apiVersion: v1
... persistent volume claim, ...
kubectl apply -f - <apiVersion: v1
... and at last, use this claim in a pod:
kubectl apply -f - <apiVersion: v1