Request for Guidance on Configuring NLB for Gearman with Auto Scaling Support

Question

We have a customer-facing application that communicates over a TCP port. Recently, we have observed Out of Memory (OOM) kills on one of the instances, and we would like to implement auto scaling based on memory metrics to improve stability and performance.

These application instances are currently behind an Application Load Balancer (ALB). After processing, the queued requests are stored in MongoDB and Redis. Presently, the entire application communicates with the queue server using its private IP over the default Gearman port 4730.

To support auto scaling and eliminate reliance on static IPs, we are planning to introduce a Network Load Balancer (NLB). This will allow us to route Gearman traffic via a DNS record (e.g., gearman.example.com) by creating an A record pointing to the NLB.

We are seeking your guidance on the complete workflow for implementing this setup. Specifically, we would like to understand:

How should we configure the NLB to support Gearman traffic over TCP port 4730?

re:Post Agent · Answer

To configure a Network Load Balancer (NLB) to support Gearman traffic over TCP port 4730 and enable auto scaling, you can follow these steps:

1. Create a target group:
   - Choose the target type as "IP" or "Instance" depending on your setup.
   - Specify the protocol as TCP and the port as 4730.
   - Configure health checks appropriate for your Gearman server.

2. Create the Network Load Balancer:
   - Select the VPC where your Gearman servers are located.
   - Choose the Availability Zones you want to enable.
   - Set the IP address type (IPv4 or dualstack).

3. Configure a listener for the NLB:
   - Set the listener protocol to TCP and the port to 4730.
   - Associate the listener with the target group you created.

4. Register targets:
   - Add your Gearman server instances or IP addresses to the target group.

5. Create an Auto Scaling group:
   - Select the appropriate launch template or configuration for your Gearman servers.
   - Choose the VPC and subnets where the instances should be launched.
   - Set the desired capacity, minimum, and maximum number of instances.
   - Configure scaling policies based on memory metrics (you may need to set up custom CloudWatch metrics for this).

6. Associate the Auto Scaling group with the target group:
   - This ensures that new instances are automatically registered with the NLB.

7. Create a DNS record:
   - In your DNS service (e.g., Route 53), create an A record for "gearman.example.com" pointing to the NLB's DNS name.

8. Update your application configuration:
   - Modify your application to use "gearman.example.com" instead of private IPs when connecting to the Gearman server.

This setup will allow your application to communicate with the Gearman servers through the NLB using the DNS name, supporting auto scaling and eliminating the need for static IPs. The NLB will distribute incoming Gearman traffic across the healthy instances in the target group, improving reliability and performance.

Remember to configure appropriate security groups to allow traffic on port 4730 between your application instances and the NLB, and from the NLB to your Gearman server instances.  
**Sources**  
[Network Load Balancers - Elastic Load Balancing](https://docs.aws.amazon.com/elasticloadbalancing/latest/network/network-load-balancers.html)  
[How to prepare architecture for 1 public elastic IP and multiple ec2 instances with routing based on port | AWS re:Post](https://repost.aws/questions/QU8JxXXYXSTBy6VVG5DgM_Kg/how-to-prepare-architecture-for-1-public-elastic-ip-and-multiple-ec2-instances-with-routing-based-on-port)

Request for Guidance on Configuring NLB for Gearman with Auto Scaling Support

Add your answer

Relevant content