Rolling Blue-Green Deployments for AWS-LB (ALB/*NLB) #1948
+792
−0
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
matt-deboer
force-pushed
the
wait_for_alb_target_group
branch
from
6a11384
to
0627fa2
Compare
matt-deboer
changed the title
matt-deboer
changed the title
matt-deboer
changed the title
matt-deboer
changed the title
matt-deboer
changed the title
radeksimko
added
enhancement
radeksimko
added
the
size/XL
|
I can't tell if there is anything left to be done on this? |
seh
reviewed
May 16, 2018
| @@ -61,6 +72,60 @@ The following attributes are exported in addition to the arguments listed above: | |||
| * `id` - The ARN of the listener (matches `arn`) | |||
| * `arn` - The ARN of the listener (matches `id`) | |||
|
|
|||
| ## Waiting for Capacity | |||
|
|
|||
| A newly-created Target Group is initially empty, with it's eventual members usually | |||
seh
reviewed
May 16, 2018
| ## Waiting for Capacity | ||
|
|
||
| A newly-created Target Group is initially empty, with it's eventual members usually | ||
| composed of ASG instances that initialize as the ASG comes up to it's desired |
seh
reviewed
May 16, 2018
| } | ||
|
|
||
| // Poll causes the poller to continuously test the output of | ||
| // it's configured url with the specified number of concurrent |
seh
reviewed
May 16, 2018
|
|
||
| errCount, err := p.waitForURLReady() | ||
| if err != nil { | ||
| return fmt.Errorf("Timed out waiting for url to become initially ready; %v", err) |
seh
reviewed
May 16, 2018
| return nil | ||
| } | ||
|
|
||
| // waitForURLReady continuously pings the configured url until |
|
|
||
| // Complete stops the poller's clients and returns | ||
| // an aggregate error describing any of the errors | ||
| // which occurred while polling the configured url |
| } | ||
|
|
||
| // rollingVersionPoller exists to continuously ping a particular | ||
| // url, testing the string result returned, and treating it the 'version' for |
| // after that, the endpoint is allowed to return one other value, considered the | ||
| // 'green' version--once the endpoint has returned a value for the 'green' version, | ||
| // it will be considered an error if the endpoint returns any other value. | ||
| // |
|
Is this PR going to be merged? Is there another way to do this without it being merged? Thanks |
|
@aeschright @seh any chance this can be looked at? I really will make my life so much better. Thanks |
|
any chance this will get merged any time soon? It's causing production errors on apply.. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds the capability of rolling (zero-downtime) deployments to the newer
aws_lbresources, similar to the rolling deployment pattern used foraws_elbresources.Even better, in this case, instances from the new ASG are available for routing until all of them are ready, and the change-over is seamless so that a given user will not received mixed-version answers from the app.
This solves #494, although the desired goal of no schema changes was not possible, due to the inherent differences in the new load-balancer resources and their decoupling from the autoscaling groups to which they route--however, if the defaults are acceptable, the 2 new optional attributes can be omitted, and the experience will be the same from the terraform user's perspective.
Changes: ( on the
aws_lb_listenerresource )Adds a
min_target_group_capacityattribute (optional); during creation and updates, the listener will wait for this number of instances to report healthy status before updating the default action. When not specified, this value is determined from the (sum of)DesiredCapacityvalue(s) on the autoscaling group(s) targeted by the target group. Set to0to disable waiting.Adds a
wait_for_capacity_timeoutattribute (optional); this configures how long the listener should wait for the desired target group capacity before failing in error. Defaults to10m.The waiting occurs on the
aws_lb_listenerresource since Target Groups only become eligible for health checks upon attachment to a Load Balancer. This approach uses a "dummy rule" (essentially non-routable) which serves to enable health checks on the target group before it is routed any requests from the load balancer.Edit (2017-10-23):
Note: NLB is not currently supported due to the lack of that capability in the provider overall (follow #1838 for details)