There are already a few deploy tools for ECS out there, but none offer deploy progress reporting for
chatops or dashboards. ecsdeploy is built as an easily deployable binary that can be fit into any
deployment pipeline, but can also be used a library to embed directly into your team's deploy tool.
ecsdeploy can report deploy progress
- As a progress bar to Slack
- As number of running/pending/stopped tasks as a webhook
- As an annotation to Grafana
Deploy status reporters to additional targets are easy to add by implementing a simple interface.
Not only that, it also has "smart" monitoring of deploys, and can determine deploy failures due to resource constraints and task failures early.
usage: ecsdeploy [<flags>] <cluster> <service>
Flags:
--help Show context-sensitive help.
--timeout=10m How long to wait for the deploy to complete
--report-addr=REPORT-ADDR URL address to report deploy status changes to
--report-auth-token=REPORT-AUTH-TOKEN
Auth token to use for reporting deploy status via HTTP. Appears on the HTTP request as
an "Authorization: Bearer <...>" header
--slack-token=SLACK-TOKEN Auth token to use for reporting deploy status to Slack
--slack-channel=SLACK-CHANNEL Slack channel to post deploy status to
--grafana-url=GRAFANA-URL URL to a Grafana server to send annotations to
--grafana-api-token=GRAFANA-API-TOKEN
API token for the Grafana server
--task-definition=TASK-DEFINITION
Location of a task definition file to deploy. If not specified, creates a new task
definition based off the currently deployed one. If "-" is specified, reads stdin.
--desired-count=-1 Desired number of tasks
--max-percent=-1 The upper limit (as a percentage of the service's desiredCount) of the number of tasks
that are allowed in the RUNNING or PENDING state in a service during a deployment.
--min-percent=-1 The lower limit (as a percentage of the service's desiredCount) of the number of
running tasks that must remain in the RUNNING state in a service during.
--tag=TAG ... Overrides the docker image tag for a container definition, written as --tag
<container_name>=<image_tag>. If there is only one container definition, the
<container_name>= prefix can be omitted. This flag can be specified multiple times to
update tags for multiple containers
--detect-failures Enable to detect deploy failures early. Failures can include: cluster not having enough
cpu/memory, image to deploy does not exist, or service is in crash loop.
Args:
<cluster> Cluster to deploy to
<service> Name of service to deploy