BlockTraffic/AllowTraffic durations

Question

Hello  
  
I've been playing with using CodeDeploy to push our applications live, but it always takes ages doing the BlockTraffic and AllowTraffic steps.  
  
The load balancer I'm using it with has connection draining set to 60s, but the block and allow traffic steps both take (reasonably reliably) 4m25 (265s) to complete which really does drag out the process.  
  
Is there a way to modify this timeout that I've missed?  Can I set the timeout on these steps in the appspec.yml even though these stages aren't documented for the hooks in that file?  
  
Thanks  
  
James  
  
Edited by: jameseogden on Apr 28, 2017 5:59 AM

Accepted Answer

Yes CodeDeploy depends on the ELB Health Check settings that you have configured your ELB with. After an instance is bound to the ELB, CodeDeploy wait for the status if the instance to be healthy ("inService") behind the load balancer. This health check is done by ELB and depends on the health check configuration you have set.

Answer

I have the same issue.  BlockTraffic and AllowTraffic take upwards of 5 minutes.  At first I thought it was due to the fact that my load balancer's Connection Draining was set to 300 seconds.  However, I disabled Connection Draining and got the same results.  I then enabled Connection Draining and set timeout to 5 seconds and still got the same results.  
  
UPDATE: CodeDeploy is dependent on the ELB Health Check settings.  My settings had been:  
  
Timeout	5 seconds  
Interval	30 seconds  
Unhealthy threshold	2  
Healthy threshold	10  
  
As a result, BlockTraffic and AllowTraffic were taking more than five minutes (30s interval * 10 healthy thresholds). After adjusting the settings to the following, BlockTraffic and AllowTraffic took about 20 seconds.  
  
Timeout	2 seconds  
Interval	5 seconds  
Unhealthy threshold	2  
Healthy threshold	2  
  
Edited by: krao411 on May 3, 2017 7:05 PM

Answer

At a deployment _CodeDeploy_ deregisters the target from the _Targetgroup_. The _Targetgroup_ now blocks new incoming traffic and gives the instance some time to finish open connections/requests. The instance is now in draining state, which is 300s by default. You can adjust it right here: **deregistration-delay**.  
  
https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-target-groups.html#deregistration-delay  
  
Cloudformation (YAML)

```
  TargetGroup:
    Type: AWS::ElasticLoadBalancingV2::TargetGroup
    Properties:
      TargetGroupAttributes:
        - Key: deregistration_delay.timeout_seconds
          Value: 60
```

That helped me to improve my deployment time.

Answer

Resolved  
  
Edited by: browner on Aug 6, 2017 4:09 AM

Answer

I'm having this exact issue and was working on this yesterday 4/27. I wonder if it's a coincidence/service issue?  
  
Currently I have a classic ELB with two EC2 nodes. So, If I do a CodeDeploy OneAtATime, the whole process takes up to 25 minutes.  
  
-m

Answer

I migrated my Classic ELB to Application... and now my BlockTraffic is taking more than 5 minutes, it used to be 1 min, health conditions are the same as before the only change is the migration to Application vs Classic ELB

Answer

I've had the same issue. Has there been any solutions to this yet?

BlockTraffic/AllowTraffic durations

相关内容