Container Configuration: Deployment
The way containers are deployed is an important step for making sure you're spreading instances to servers with available resources, getting copies of your container onto several servers in different data centers, or, with manual deployments, telling Cycle exactly where you'd like your deployments to go.
Initial Instances
Initial instances refers to the number of instances to create during the container create process. That initial number will be seen by Cycle and then according to your deployment strategy, the instances will be created on the servers that best match available resources.
Deployment Strategy
Get full info on deployment strategies here
Function
For function containers, there is additional settings available:
- Max Pool Size - The maximum number of instances that Cycle can pre-allocate (includes auto-scaled instances).
- Max Shard Concurrency - For each shard (scheduler), the maximum number of tasks it can run in parallel.
- Max Runtime - The maximum amount of time a function instance can run before timing out.
- Max Queue Time - The maximum amount of time Cycle will wait for an instance to be available.
Constraints
The constraints form has two way to tag container instances:
Tags Any - A list of tags for the instance. If a server matches any tag in the list - that server becomes a viable candidate for the instances.
Tags All - A list of tags for the instance. A server must match every tag on the list in order to become a viable candidate for the instances.
Startup Policy
A delay, in a time string, that Cycle will wait before sending the start signal to this container upon environment start.
This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
These settings are related to how the container updates its image.
Shutdown Policy
Information on how Cycle should handle a shutdown signal.
Graceful Timeout - A time string the platform will wait for a container to stop gracefully. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
Restart Policy
Information about how Cycle should handle a restart event.
Restart Condition - If the container fails under what conditions should Cycle attempt a restart.
Delay - A time string for how long to wait between restart attempts. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
Max Restart - The maximum number of times to try and restart this container.
Health Check Policy
Commands and instructions Cycle will run to verify your containers health.
Command - The command you want Cycle to run, success will mean that the container is healthy, and error will mean that it is not.
If you want to describe an "inline script" in the command form field use "" quotes to wrap the script. Writing something like sh -c "script here" will not work correctly because of the way the sh shell exits.
Retries - How many times to try the command.
Delay - How long to wait before performing an initial health check when the instance starts. The state of the instance will be `null`` until the first check is performed (see below for more info).
Interval - A time string between tries to wait before trying again. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
Timeout - A time string that says the amount of time to wait before the assumption the the command has failed. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
Restart on Fail - If checked, when the container enters the unhealthy state it will automatically restart.
To learn more about the health state of the container instances as they respond to health checks check the Instances API Docs. Here users can expand the data object, then state, and finally health to get an idea of what the health states mean.
More advanced uses may use this information to build a healthiness probe for their instances.
Telemetry
Settings for how you'd like Cycle to collect your telemetry data.
Retention - A time string for how long the platform will retain telemetry for a given instance. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
Interval - A time string that represents how much time your would like each telemetry report to cover. This accepts a string with modifiers. Example for 3 days 15 hours 10 minutes 5 seconds 3d15h10m5s
WebHook - a url that Cycle will post your telemetry reports to.
Disable Telemetry - if checked, this will disable all telemetry reporting for this container.
Stateful Instances
There are a special class of configuration settings specifically for stateful instances. These settings will only be present when a stateful instance is being configured and will not show at all in stateless instance configuration panels.
Use Base Hostname
For containers that need to use the "base hostname" or the hostname without the instance number signifier prepended to it (ex 1.hostname).
Update
This setting is used to configure behavior of the container instances on "update" (reimage). The main setting is "Stagger", which, when set, will cause the platform to pick a random time from 0 - this duration. This stagger is applied to the instances so they all re-start at different times (up to the time specified here).
When reimaging a container, the Cycle platform will fully download the image to the server before the reimage step is executed.