How do stage conditions work? - Amazon CodePipeline
Rules for stage conditions
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

How do stage conditions work?

For each condition that specifies a rule, the rule is run. If the condition fails, the result is engaged. The stage performs the specified result only when the condition fails. Optionally, as part of the rule, you also specify which resources CodePipeline should use for certain cases. For example, the CloudWatchAlarm rule will use a CloudWatch alarm resource to run checks for the condition.

A condition might match multiple rules, and each rule can specify one of three providers.

The high-level flow for creating conditions as as follows.

  1. Choose the type of condition from the available condition types in CodePipeline. For example, use an On Success condition type to set up a stage so that after the stage succeeds, a set of rules can be used to run checks before proceeding.

  2. Choose the rule. For example, the CloudWatchAlarm rule will check for alarms and uses EB to check for a preconfigured alarm threshold. If the check is successful, and the alarm is below the threshold, the stage can proceed.

  3. Configure the result, such as a rollback that would be used if the rule fails.

There are three types of condition you can configure, and there are valid result options for each type of condition as follows.

  1. Result option for Entry conditions: fail

  2. Result option for On Failure conditions: rollback

  3. Result options for On Success conditions are either rollback or fail.

The following diagram shows an example flow for the Entry condition type in CodePipeline. Conditions answer the question, What should happen if the condition is not met, meaning any rule fails? In the following flow, an Entry condition is configured with a LambdaInvoke rule and a CloudWatchAlarm rule. If the rule fails, then the configured result, such as Fail, is engaged.

An example of the Entry condition type with two rules configured, a LambdaInvoke rule and a CloudWatchAlarm rule.

The following diagram shows an example flow for the On Failure condition type in CodePipeline. Conditions answer the question, What should happen if the condition is met, meaning the rules all succeed their checks? In the following flow, an On Failure condition is configured with a LambdaInvoke rule and a CloudWatchAlarm rule. If the rule succeeds, then the configured result, such as Fail, is engaged.

An example of the On Failure condition type with two rules configured, a Lambda rule and a CloudWatchAlarm rule.

The following diagram shows an example flow for the On Success condition type in CodePipeline. Conditions answer the question, What should happen if the condition is met, meaning the rules all succeed their checks? In the following flow, an On Success condition is configured with a LambdaInvoke rule and a CloudWatchAlarm rule. If the rule succeeds, then the configured result, such as Fail, is engaged.

An example of the On Success condition type with two rules configured, a Lambda rule and a CloudWatchAlarm rule.

Rules for stage conditions

When you configure stage conditions, you select from pre-defined rules and specify the results for the rule. A condition state will be Failed if any of the rules in the condition are failed and Succeeded if all the rules succeed. How criteria are met for conditions for On Failure and On Success depends on the type of rule.

The following are managed rules that you can add to stage conditions.

  • Conditions can use the Amazon DeploymentWindow rule to specify approved deployment times for allowing a deployment. The criteria for the rule will be measured with a provided cron expression for a deployment window. The rule succeeds when the date and time in the deployment window meet the criteria in the cron expression for the rule. For more information about this rule, see DeploymentWindow.

  • Conditions can use the Amazon Lambda rule to check for error states returned from configured Lambda functions. The rule is met when the check receives the Lambda function result. An error from the Lambda function meets the criteria for On Failure conditions. For more information about this rule, see LambdaInvoke.

  • Conditions can use the Amazon CloudWatchAlarm rule to check for alarms configured from CloudWatch events. The rule is met when the check returns an alarm state of OK, ALARM, or INSUFF_DATA. For On Success conditions, OK and INSUFFICIENT_DATA meet the criteria. ALARM meets the criteria for On Failure conditions. For more information about this rule, see CloudWatchAlarm.