Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Containers guide #3347

Merged
merged 9 commits into from
Aug 1, 2017
Merged

Containers guide #3347

merged 9 commits into from
Aug 1, 2017

Conversation

preetapan
Copy link
Member

Addressed a bunch of review comments, pls take a final look.

## Consul Official Docker Image
Consul's official Docker images are tagged with version numbers. For example, `docker pull consul:0.9.0` will pull the 0.9.0 Consul release image.

For major releases, make sure to read our [upgrade guides](/docs/upgrade-specific.html) before upgrading your cluster.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

  • s/our/the
  • s/your/a

As a general rule, try to avoid first/second person in technical documentation where possible.


For major releases, make sure to read our [upgrade guides](/docs/upgrade-specific.html) before upgrading your cluster.

To get a development mode consul instance running the latest version, run `docker run consul`.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Capital "C" consul here


## Data Directory Persistence

The container exposes its data directory, `/consul/data`, as a [volume](https://docs.docker.com/engine/tutorials/dockervolumes/). This is where Consul will place its persisted state.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/place/store


Configuration can also be added by passing the configuration JSON via environment variable CONSUL_LOCAL_CONFIG. Example:

```
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we:

  1. Add sh after the ticks to add syntax highlighting
  2. Start the command with a $ to indicate it's a command
  3. Left-flush this (it's indented very far right now

```

## Networking
When run inside a container, Consul's IP addresses need to be setup properly. Consul has to be configured with an appropriate _cluster address_ as well as a _client address._ In some cases, it might also require configuring an _advertise_address_.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we combine this into:

When running inside a container, Consul must be configured with an appropriate cluster address and client address. In some cases, it may also require configuring an advertise address.

1. Environment Variables: Use the `CONSUL_CLIENT_INTERFACE` and `CONSUL_BIND_INTERFACE` environment variables. In the following example `eth0` is the network interface of the container.

```
docker run \
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Left align to the "```" + sh

In the following example, the client and bind addresses are declaratively specified for the container network interface 'eth0'

```
docker run \
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Left align to the "```" + sh

```

## Stopping and Restarting Containers
Consul containers can be stopped using the `docker stop <container_id>` command and restarted using `docker start <container_id>`. As long as there are enough servers in the cluster to maintain [quorum](/docs/internals/consensus.html#deployment-table), Consul's [Autopilot](/docs/guides/autopilot.html) feature will handle removing servers whose containers were stopped. Autopilot's default settings are already configured correctly. If you override them, make sure that the following [settings](/docs/agent/options.html#autopilot) are appropriate.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's usually helpful to callout commands outside of the text. Consider:

The official Consul container supports stopping, starting, and restarting. To stop a container, use docker stop:

$ docker stop <container-id>

To start a container, run docker start:

$ docker start <container-id>

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it worth mentioning docker kill here?

$ docker kill --signal="HUP" <container name or id>

That would also do an in-memory reload.

* `last_contact_threshold` should be reasonably small, so that dead servers are removed quickly.
* `server_stabilization_time`should be sufficiently large( on the order of several seconds) so that unstable servers are not added to the cluster until they stabilize.

If a container that was running a leader is stopped, leader election will be triggered causing another server in the cluster to assume leadership.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If the container running the currently-elected Consul server leader is stopped, a leader election will trigger. This event will cause a new Consul server in the cluster to assume leadership.

## Known Issues
**Consul does not currently gracefully handle the situation where all nodes in the cluster running inside a container are restarted at the same time, and they all obtain new IP addresses.** This is because the underlying Raft layer persists the IP address and needs it for leader election operations. Operators must carefully orchestrate restarts of Consul containers that have ephemeral IP addresses to do restarts in small numbers, so that they can gracefully leave the cluster and re-join with their new IP address.

**Snapshot close error** Due to a [known issue](https://github.com/docker/libnetwork/issues/1204) with half close support in Docker, you will see an error message `[ERR] consul: Failed to close snapshot: write tcp <source>-><destination>: write: broken pipe` when saving snapshots. This doesn't affect saving and restoring snapshots when running in Docker.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/doesn't/does not


The container exposes its data directory, `/consul/data`, as a [volume](https://docs.docker.com/engine/tutorials/dockervolumes/). This is where Consul will store its persisted state.

For clients, this stores some information about the cluster and the client's health checks in case the container is restarted. If the volume on a client disappears, it doesn't affect cluster operations.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'd say "services and health checks"


* **Cluster Address** - The address at which other Consul agents may contact a given agent. This is also referred to as the bind address.

* **Client Address** - The address where other processes on the host contact Consul in order to make HTTP or DNS requests. Consider setting this to localhost or `127.0.0.1` to only allow processes on the same container to make http/DNS requests.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s/http/HTTP/


* **Client Address** - The address where other processes on the host contact Consul in order to make HTTP or DNS requests. Consider setting this to localhost or `127.0.0.1` to only allow processes on the same container to make http/DNS requests.

* **Advertise Address** - The advertise address is used to change the address that we advertise to other nodes in the cluster. This defaults to the bind address. Consider using this if you use NAT tables in your environment, or in scenarios where you have a routable address that cannot be bound
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

delete "tables"

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Need a period at the end of this sentence.


* `cleanup_dead_servers` must be set to true to make sure that a stopped container is removed from the cluster.
* `last_contact_threshold` should be reasonably small, so that dead servers are removed quickly.
* `server_stabilization_time`should be sufficiently large( on the order of several seconds) so that unstable servers are not added to the cluster until they stabilize.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

need a space before "should"

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also space after large, and there's an extra space inside the parens.

Copy link
Contributor

@slackpad slackpad left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Couple tiny things, then LGTM.

@preetapan preetapan merged commit 5ca685a into master Aug 1, 2017
@slackpad slackpad deleted the containers_guide branch August 8, 2017 18:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

4 participants