Skip to content

Commit

Permalink
Merge pull request kubernetes#18053 from amygdala/docs
Browse files Browse the repository at this point in the history 
Polishing the Deployment doc
  • Loading branch information
Marek Grabowski committed Dec 3, 2015
2 parents 5d92cb2 + 3f2e5ed commit 80415f2
Showing 1 changed file with 75 additions and 64 deletions.
139 changes: 75 additions & 64 deletions docs/user-guide/deployments.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@ Documentation for other releases can be found at

- [Deployments](#deployments)
- [What is a _Deployment_?](#what-is-a-deployment)
- [Enabling Deployments on kubernetes cluster](#enabling-deployments-on-kubernetes-cluster)
- [Enabling Deployments on a Kubernetes cluster](#enabling-deployments-on-a-kubernetes-cluster)
- [Creating a Deployment](#creating-a-deployment)
- [Updating a Deployment](#updating-a-deployment)
- [Multiple Updates](#multiple-updates)
Expand All @@ -60,22 +60,23 @@ Documentation for other releases can be found at

## What is a _Deployment_?

A _Deployment_ provides declarative update for Pods and ReplicationControllers.
Users describe the desired state in deployment object and deployment
controller changes the actual state to that at a controlled rate.
Users can define deployments to create new resources, or replace existing ones
A _Deployment_ provides declarative updates for Pods and ReplicationControllers.
Users describe the desired state in a Deployment object, and the deployment
controller changes the actual state to the desired state at a controlled rate.
Users can define Deployments to create new resources, or replace existing ones
by new ones.

A typical use case is:
* Create a deployment to bring up a replication controller and pods.
* Later, update that deployment to recreate the pods (for ex: to use a new image).
* Create a Deployment to bring up a replication controller and pods.
* Later, update that Deployment to recreate the pods (for example, to use a new image).

## Enabling Deployments on kubernetes cluster
## Enabling Deployments on a Kubernetes cluster

Deployments is part of the [`extensions` API Group](../api.md#api-groups) and is not enabled by default.
Set `--runtime-config=extensions/v1beta1/deployments=true` on API server to
Deployment objects are part of the [`extensions` API Group](../api.md#api-groups) and this feature
is not enabled by default.
Set `--runtime-config=extensions/v1beta1/deployments=true` on the API server to
enable it.
This can be achieved by exporting `ENABLE_DEPLOYMENTS=true` before running
This can be achieved by exporting `KUBE_ENABLE_DEPLOYMENTS=true` before running the
`kube-up.sh` script on GCE.

Note that Deployment objects effectively have [API version
Expand Down Expand Up @@ -120,26 +121,31 @@ $ kubectl create -f docs/user-guide/nginx-deployment.yaml
deployment "nginx-deployment" created
```

Running a get immediately will give:
Running

```console
$ kubectl get deployments
```

immediately will give:

```console
$ kubectl get deployments
NAME UPDATEDREPLICAS AGE
nginx-deployment 0/3 8s
```

This indicates that deployment is trying to update 3 replicas. It has not
updated any one of those yet.
This indicates that the Deployment is trying to update 3 replicas, and has not updated any of them yet.

Running a get again after a minute, will give:
Running the `get` again after a minute, should give:

```console
$ kubectl get deployments
NAME UPDATEDREPLICAS AGE
nginx-deployment 3/3 1m
```

This indicates that deployent has created all the 3 replicas.
This indicates that the Deployment has created all three replicas.
Running ```kubectl get rc``` and ```kubectl get pods``` will show the replication controller (RC) and pods created.

```console
Expand All @@ -157,13 +163,13 @@ deploymentrc-1975012602-j975u 1/1 Running 0 1m
deploymentrc-1975012602-uashb 1/1 Running 0 1m
```

The created RC will ensure that there are 3 nginx pods at all time.
The created RC will ensure that there are three nginx pods at all times.

## Updating a Deployment

Lets say, now we want to update the nginx pods to start using nginx:1.9.1 image
instead of nginx:1.7.9.
For this, we update our deployment to be as follows:
Suppose that we now want to update the nginx pods to start using the `nginx:1.9.1` image
instead of the `nginx:1.7.9` image.
For this, we update our deployment file as follows:

<!-- BEGIN MUNGE: EXAMPLE new-nginx-deployment.yaml -->

Expand All @@ -189,13 +195,14 @@ spec:
[Download example](new-nginx-deployment.yaml?raw=true)
<!-- END MUNGE: EXAMPLE new-nginx-deployment.yaml -->
We can then `apply` the Deployment:

```console
$ kubectl apply -f docs/user-guide/new-nginx-deployment.yaml
deployment "nginx-deployment" configured
```

Running a get immediately will still give:
Running a `get` immediately will still give:

```console
$ kubectl get deployments
Expand All @@ -205,26 +212,26 @@ nginx-deployment 3/3 8s

This indicates that deployment status has not been updated yet (it is still
showing old status).
Running a get again after a minute, will give:
Running a `get` again after a minute, should show:

```console
$ kubectl get deployments
NAME UPDATEDREPLICAS AGE
nginx-deployment 1/3 1m
```

This indicates that deployment has updated one of the three pods that it needs
This indicates that the Deployment has updated one of the three pods that it needs
to update.
Eventually, it will get around to updating all the pods.
Eventually, it will update all the pods.

```console
$ kubectl get deployments
NAME UPDATEDREPLICAS AGE
nginx-deployment 3/3 3m
```

We can run ```kubectl get rc``` to see that deployment updated the pods by creating a new RC
which it scaled up to 3 and scaled down the old RC to 0.
We can run ```kubectl get rc``` to see that the Deployment updated the pods by creating a new RC,
which it scaled up to 3 replicas, and has scaled down the old RC to 0 replicas.

```console
kubectl get rc
Expand All @@ -233,7 +240,7 @@ deploymentrc-1562004724 nginx nginx:1.9.1 deployment.kubernetes.io/
deploymentrc-1975012602 nginx nginx:1.7.9 deployment.kubernetes.io/podTemplateHash=1975012602,app=nginx 0 7m
```

Running get pods, will only show the new pods.
Running `get pods` should now show only the new pods:

```console
kubectl get pods
Expand All @@ -243,7 +250,7 @@ deploymentrc-1562004724-1rkfl 1/1 Running 0 8m
deploymentrc-1562004724-6v702 1/1 Running 0 8m
```

Next time we want to update pods, we can just update the deployment again.
Next time we want to update these pods, we can just update and re-apply the Deployment again.

Deployment ensures that not all pods are down while they are being updated. By
default, it ensures that minimum of 1 less than the desired number of pods are
Expand Down Expand Up @@ -273,27 +280,29 @@ Events:
1m 1m 1 {deployment-controller } ScalingRC Scaled down rc deploymentrc-1975012602 to 0
```

Here we see that when we first created the deployment, it created an RC and scaled it up to 3 replicas directly.
When we updated the deployment, it created a new RC and scaled it up to 1 and then scaled down the old RC by 1, so that at least 2 pods were available at all times.
Here we see that when we first created the Deployment, it created an RC and scaled it up to 3 replicas directly.
When we updated the Deployment, it created a new RC and scaled it up to 1 and then scaled down the old RC by 1, so that at least 2 pods were available at all times.
It then scaled up the new RC to 3 and when those pods were ready, it scaled down the old RC to 0.

### Multiple Updates

Each time a new deployment object is observed, a replication controller is
created to bring up the desired pods if there is no existing RC doing so.
Existing RCs controlling pods whose labels match `.spec.selector` but the
Existing RCs controlling pods whose labels match `.spec.selector` but whose
template does not match `.spec.template` are scaled down.
Eventually, the new RC will be scaled to `.spec.replicas` and all old RCs will
be scaled to 0.
If the user updates the deployment while an existing deployment was in progress,
deployment will create a new RC as per the update and start scaling that up and
will roll the RC that it was scaling up before in its list of old RCs and will

If the user updates a Deployment while an existing deployment is in progress,
the Deployment will create a new RC as per the update and start scaling that up, and
will roll the RC that it was scaling up previously-- it will add it to its list of old RCs and will
start scaling it down.
For example: If user creates a deployment to create 5 replicas of nginx:1.7.9.
But then updates the deployment to create 5 replicas of nging:1.9.1, when only 3
replicas of nginx:1.7.9 had been created, then deployment will immediately start
killing the 3 nginx:1.7.9 pods that it had created and will start creating
nginx:1.9.1 pods. It will not wait for 5 replicas of nginx:1.7.9 to be created

For example, suppose the user creates a deployment to create 5 replicas of `nginx:1.7.9`,
but then updates the deployment to create 5 replicas of `nginx:1.9.1`, when only 3
replicas of `nginx:1.7.9` had been created. In that case, deployment will immediately start
killing the 3 `nginx:1.7.9` pods that it had created, and will start creating
`nginx:1.9.1` pods. It will not wait for 5 replicas of `nginx:1.7.9` to be created
before changing course.

## Writing a Deployment Spec
Expand All @@ -309,12 +318,12 @@ A Deployment also needs a [`.spec` section](../devel/api-conventions.md#spec-and
The `.spec.template` is the only required field of the `.spec`.

The `.spec.template` is a [pod template](replication-controller.md#pod-template). It has exactly
the same schema as a [pod](pods.md), except it is nested and does not have an
the same schema as a [pod](pods.md), except that it is nested and does not have an
`apiVersion` or `kind`.

### Replicas

`.spec.replicas` is an optional field that specifies the number of desired pods. Defaults
`.spec.replicas` is an optional field that specifies the number of desired pods. It defaults
to 1.

### Selector
Expand All @@ -331,26 +340,26 @@ number of pods are less than the desired number.
is added to existing RCs (and label key that is added to its pods) to prevent
the existing RCs to select new pods (and old pods being selected by new RC).
Users can set this to an empty string to indicate that the system should
not add any selector and label. If unspecified, system uses
"deployment.kubernetes.io/podTemplateHash".
Value of this key is hash of `.spec.template`.
No label is added if this is set to empty string.
not add any selector and label. If unspecified, the system uses
`deployment.kubernetes.io/podTemplateHash`.
The value of this key is the hash of `.spec.template`.
No label is added if this is set to the empty string.

### Strategy

`.spec.strategy` specifies the strategy to replace old pods by new ones.
`.spec.strategy` specifies the strategy used to replace old pods by new ones.
`.spec.strategy.type` can be "Recreate" or "RollingUpdate". "RollingUpdate" is
the default value.

#### Recreate Deployment

All existing pods are killed before new ones are created when
`.spec.strategy.type==Recreate`.
Note: This is not implemented yet.
__Note: This is not implemented yet__.

#### Rolling Update Deployment

Deployment updates pods in a [rolling update](update-demo/) fashion
The Deployment updates pods in a [rolling update](update-demo/) fashion
when `.spec.strategy.type==RollingUpdate`.
Users can specify `maxUnavailable`, `maxSurge` and `minReadySeconds` to control
the rolling update process.
Expand All @@ -359,46 +368,48 @@ the rolling update process.

`.spec.strategy.rollingUpdate.maxUnavailable` is an optional field that specifies the
maximum number of pods that can be unavailable during the update process.
Value can be an absolute number (ex: 5) or a percentage of desired pods (ex:
10%).
Absolute number is calculated from percentage by rounding up.
The value can be an absolute number (e.g. 5) or a percentage of desired pods
(e.g. 10%).
The absolute number is calculated from percentage by rounding up.
This can not be 0 if `.spec.strategy.rollingUpdate.maxSurge` is 0.
By default, a fixed value of 1 is used.
Example: when this is set to 30%, the old RC can be scaled down to

For example, when this value is set to 30%, the old RC can be scaled down to
70% of desired pods immediately when the rolling update starts. Once new pods are
ready, old RC can be scaled down further, followed by scaling up the new RC,
ensuring that the total number of pods available at all times during the
update is at least 70% of desired pods.
update is at least 70% of the desired pods.

##### Max Surge

`.spec.strategy.rollingUpdate.maxSurge` is an optional field that specifies the
maximum number of pods that can be created above the desired number of pods.
Value can be an absolute number (ex: 5) or a percentage of desired pods (ex:
10%).
This can not be 0 if MaxUnavailable is 0.
Absolute number is calculated from percentage by rounding up.
Value can be an absolute number (e.g. 5) or a percentage of desired pods
(e.g. 10%).
This can not be 0 if `MaxUnavailable` is 0.
The absolute number is calculated from percentage by rounding up.
By default, a value of 1 is used.
Example: when this is set to 30%, the new RC can be scaled up immediately when

For example, when this value is set to 30%, the new RC can be scaled up immediately when
the rolling update starts, such that the total number of old and new pods do not exceed
130% of desired pods. Once old pods have been killed,
new RC can be scaled up further, ensuring that total number of pods running
at any time during the update is atmost 130% of desired pods.
the new RC can be scaled up further, ensuring that the total number of pods running
at any time during the update is at most 130% of desired pods.

##### Min Ready Seconds

`.spec.strategy.rollingUpdate.minReadySeconds` is an optional field that specifies the
minimum number of seconds for which a newly created pod should be ready
without any of its container crashing, for it to be considered available.
Defaults to 0 (pod will be considered available as soon as it is ready).
Note: This is not implemented yet.
without any of its containers crashing, for it to be considered available.
This defaults to 0 (the pod will be considered available as soon as it is ready).
__Note: This is not implemented yet__.

## Alternative to Deployments

### kubectl rolling update

[Kubectl rolling update](kubectl/kubectl_rolling-update.md) also updates pods and replication controllers in a similar fashion.
But deployments is declarative and is server side.
But Deployments is declarative and is server side.


<!-- BEGIN MUNGE: GENERATED_ANALYTICS -->
Expand Down

0 comments on commit 80415f2

Please sign in to comment.