CouchDB 2.0 for Kubernetes w/ manifests

Maintainer

Description

Minimal image with a sidecar container that performs automatic cluster initialization. There is also support for local dev environments to auto-finalize the cluster upon starting, as well as a couch-helper utility script. This image uses a custom version of Debian Linux (Jessie) that I designed weighing in at ~22MB compressed.

Build Environment

Build environment variables are often used in the build script to bump version numbers and set other options during the docker build phase. Their values can be overridden using a build argument of the same name.

ERLANG_VERSION
COUCHDB_VERSION

The following variables are standard in most of our dockerfiles to reduce duplication and make scripts reusable among different projects:

APP: couchdb
USER: couchdb
HOME /opt/couchdb

Run Environment

Run environment variables are used in the entrypoint script to render configuration templates, perform flow control, etc. These values can be overridden when inheriting from the base dockerfile, specified during docker run, or in kubernetes manifests in the env array.

couchdb container:

ERLANG_THREADS: used as the value for the +A argument in vm.args. Defaults to 25.
COUCHDB_LOG_LEVEL: lowercased and used as the value for the level in the log section of local.ini. Defaults to info.
COUCHDB_DATA_PATH: used as the value for database_dir and view_index_dir in the couchdb section of local.ini. Defaults to /data/$APP.
COUCHDB_BIND_ADDR: used as the value for bind_address in the chttpd and httpd sections of local.ini. Defaults to 0.0.0.0.
COUCHDB_REQUIRE_VALID_USER: used as the value for require_valid_user in the chttpd and httpd sections of local.ini. Defaults to false.
COUCHDB_SHARDS: used as the value for q in the cluster section of local.ini. Defaults to 4.
COUCHDB_READ_QUORUM: used as the value for r in the cluster section of local.ini. Defaults to 1.
COUCHDB_WRITE_QUORUM: used as the value for w in the cluster section of local.ini. Defaults to 1.
COUCHDB_REPLICAS: used as the value for n in the cluster section of local.ini. Defaults to 1.
LOCAL_DEV_CLUSTER: when value is true, will trigger the couch-helper in the entrypoint script before starting couchdb. Inject as true into the environment when running a simple single node dev cluster that should immediately auto-initialize. Defaults to false.
COUCHDB_ADMIN_USER,COUCHDB_ADMIN_PASS: when set this will be available to the sidecar container couchdiscover and the script couch-helper. Your cluster will be auto initialized using these credentials. couch-helper is meant to be used for local single node clusters for development.
ERLANG_COOKIE: when set this value will be written to ~/.erlang.cookie and proper permissions applied prior to starting couchdb.
COUCHDB_CLUSTER_SIZE: when set this value override the value of the replica's field on the kubernetes statefulset manifest. Do not use unless you really need to override the default behavior for some reason.

couchdiscover container:

LOG_LEVEL: logging level to output container logs for. Defaults to INFO, most logs are either INFO or WARNING level.

Usage

Under docker (pre-built)

All of our docker-* repos in github have CI pipelines that push to docker cloud/hub.

This image is available at:

https://store.docker.com/community/images/callforamerica/couchdb
https://hub.docker.com/r/callforamerica/couchdb.
docker pull callforamerica/couchdb

To run:

docker run -d \
    --name couchdb \
    -h couchdb.local \
    -e "LOCAL_DEV_CLUSTER=true" \
    -e "COUCHDB_ADMIN_USER=admin" \
    -e "COUCHDB_ADMIN_PASS=secret" \
    -e "ERLANG_COOKIE=test-cookie" \
    callforamerica/couchdb

NOTE: Please reference the Run Environment section for the list of available environment variables.

Under docker-compose

Pull the images

docker-compose pull

Start application and dependencies

# start in foreground
docker-compose up --abort-on-container-exit

# start in background
docker-compose up -d

Under Kubernetes

Edit the manifests under kubernetes/<environment> to reflect your specific environment and configuration.

Create a secret for the erlang cookie:

kubectl create secret generic erlang-cookie --from-literal=erlang.cookie=$(LC_ALL=C tr -cd '[:alnum:]' < /dev/urandom | head -c 64)

Create a secret for the couchdb credentials:

kubectl create secret generic couchdb-creds --from-literal=couchdb.user=$(sed $(perl -e "print int rand(99999)")"q;d" /usr/share/dict/words) --from-literal=couchdb.pass=$(LC_ALL=C tr -cd '[:alnum:]' < /dev/urandom | head -c 32)

Deploy couchdb:

kubectl create -f kubernetes/<environment>

couchdiscover

For the kubernetes manifests, there is a sidecar container called couchdiscover that handles initializing the cluster.

In order to best use something that is essentially "zero configuration," it helps to understand how the necessary information is obtained from the environment and api's.

How couchdiscover discovers information

Initially a great deal of information is obtained by grabbing the hostname of the container that's part of a StatefulSet and parsing it. This is how the namespace is determined, how hostnames are calculated later, the name of the StatefulSet to look for in the api, the name of the headless service, the node name, the index, whether a node is master or not, etc.
The kubernetes api is used to grab the StatefulSet and Endpoint objects. The Endpoint object is parsed to obtain the hosts list. Then the StatefulSet is parsed for the ports, then the environment is resolved, fetching any externally referenced ConfigMaps or Secrets that are necessary. Credentials are resolved by looking through the environment of the couchdb container for the keys: COUCHDB_ADMIN_USER, COUCHDB_ADMIN_PASS. Finally the expected cluster size is set to the number of replicas in the fetched StatefulSet. You can override this by setting this environment variable manually, but should be completely unnecessary for most cases.

Main logic

The main logic is performed in the manage module's ClusterManager object's run method. I think most of it is relatively straightforward.

# couchdiscover.manage.ClusterManager
def run(self):
    """Main logic here, this is where we begin once all environment
    information has been retrieved."""
    log.info('Starting couchdiscover: %s', self.couch)
    if self.couch.disabled:
        log.info('Cluster disabled, enabling')
        self.couch.enable()
    elif self.couch.finished:
        log.info('Cluster already finished')
        self.sleep_forever()

    if self.env.first_node:
        log.info("Looks like I'm the first node")
        if self.env.single_node_cluster:
            log.info('Single node cluster detected')
            self.couch.finish()
    else:
        log.info("Looks like I'm not the first node")
        self.couch.add_to_master()
        if self.env.last_node:
            log.info("Looks like I'm the last node")
            self.couch.finish()
        else:
            log.info("Looks like I'm not the last node")
    self.sleep_forever()

haibui-htk/docker-couchdb