Listing all the pods belonging to a deployment can be done by querying its selectors, but using the deployment’s synthesized replicaset identifier allows for easier automation.
# deployment name and namespace
deployment_name=mydeployment
deployment_ns=mynamespace
# get replica set identifier for deployment
dep_rs=$(kubectl describe deployment $deployment_name -n $deployment_ns | grep ^NewReplicaSet | awk '{print $2}')
# get pod hash for replicaset
pod_hash=$(kubectl get replicaset $dep_rs -n $deployment_ns -o jsonpath="{.metadata.labels.pod-template-hash}")
# show all pods belonging to deployment
kubectl get pods -l pod-template-hash=$pod_hash -n $deployment_ns
The idea of pulling the ‘NewReplicateSet’ identifier comes from this page by 9to5answer. For a more general solution based on ‘ownerReferences’, see the section below.
(Alternative) Pulling all pods from Deployment using ownerReferences
The solution above has the advantage of not requiring any 3rd party utilities, but if we use the ‘jq‘ utility it provides enhanced filtering expressions that give us a more general solution based on pulling pods via their ‘ownerReferences’.
# deployment name and namespace deployment_name=mydeployment deployment_ns=mynamespace # get sythesized replica set identifier for deployment dep_rs=$(kubectl get replicaset -n $deployment_ns -o=json | jq -r ".items[].metadata | select(.ownerReferences[0].name==\"$deployment_name\" and .ownerReferences[0].kind==\"Deployment\") | .name") # show all pods controlled by Deployment (via synthesized ReplicaSet) kubectl get pods -n $deployment_ns -o=json | jq -r ".items[].metadata | select(.ownerReferences[0].name==\"$dep_rs\" and .ownerReferences[0].kind==\"ReplicaSet\") | .name"
Pulling all pods from ReplicaSet using ownerReferences
Using the same ‘ownerReferences’ field as the example above, we can pull all the pods controlled by a standard ReplicaSet.
# deployment name and namespace replicaset_name=myreplicaset replicaset_ns=mynamespace # show all pods controlled by ReplicaSet kubectl get pods -n $replicaset_ns -o=json | jq -r ".items[].metadata | select(.ownerReferences[0].name==\"$replicaset_name\" and .ownerReferences[0].kind==\"ReplicaSet\") | .name"
Pulling all pods from DaemonSet using ownerReferences
Using the same ‘ownerReferences’ field as the example above, we can pull all the pods controlled by a DaemonSet.
# deployment name and namespace damonset_name=mydaemonset daemonset_ns=mynamespace # show all pods controlled by DaemonSet kubectl get pods -n $daemonset_ns -o=json | jq -r ".items[].metadata | select(.ownerReferences[0].name==\"$daemonset_name\" and .ownerReferences[0].kind==\"DaemonSet\") | .name"
REFERENCES
9to5answer.com, list pods of deployment using NewReplicaSet identifier
NOTES
It may seem like parsing the describe for ‘NewReplicaSet’ smells, but pulling the data from get is not any better.
# curently using this on 'describe'
kubectl describe deployment $deployment_name -n $deployment_ns | grep ^NewReplicaSet | awk '{print $2}'
# because this does not look easier
kubectl get replicaset -n $deployment_ns -o=custom-columns="NAME:.metadata.name,POD:.metadata.ownerReferences[?(@.name=='$deployment_name')].name" --no-headers | grep -v '' | awk {'print $1'}
# and this is messier
kubectl get deployment $deployment_name -n $deployment_ns -o jsonpath='{.status.conditions[?(@.type=="Progressing")].message}' | grep -Po "\"([^\"]*)\"" | tr -d '"'
The jsonpath support in kubectl is not sufficient to put complex expressions in either custom-columns or field-selector
controlled_by_name="abc" # unfortunate that AND|OR operations not supported in kubectl # or we could add multiple conditions and remove need for jq kubectl get pods -n default -o=custom-columns="POD:.metadata.ownerReferences[?(@.name=='$controlled_by_name')].name"