Hey there, let's troubleshoot this frustrating single-user access issue with your Dockerized Tomcat on AWS. Since everything works perfectly without Docker, the problem is definitely tied to how your container is set up or deployed. Here are the most likely causes and fixes to try out:

Tomcat's default connector settings might be too restrictive inside your container, especially around concurrent connections. Pop open your server.xml file and look at the Connector configuration—here's what a reasonable setup looks like:

<Connector port="8080" protocol="HTTP/1.1"
           connectionTimeout="20000"
           maxThreads="200"
           acceptCount="100"
           maxConnections="1000" />

• maxThreads: This is the maximum number of threads Tomcat uses to handle requests. If this was cranked down in your container config, it'll cap concurrent users. Match it to what you use in your non-Docker Tomcat setup.
• acceptCount: When all threads are busy, this is the max number of requests that can wait in the queue. Too small, and new requests get rejected immediately.

AWS containers often have CPU/memory limits that can strangle Tomcat's ability to handle multiple users.

• Check your deployment config (whether it's docker-compose.yml or an AWS ECS task definition) for resource limits like cpus or mem_limit. For example:

services:
  tomcat:
    image: your-custom-tomcat-image
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 512M

If you see something like that, bump up the resources—try 1 CPU and 1GB of memory first. Tomcat needs enough headroom to spin up threads for concurrent users.

If you're using an AWS Load Balancer (ALB/NLB) to expose your container, misconfigurations here could be blocking concurrent access:

• Double-check port mapping: Make sure your container's Tomcat port (usually 8080) is correctly mapped to the host/load balancer port—no conflicts, no weird single-connection restrictions.
• Check load balancer timeouts: An overly short idle timeout might kill connections prematurely. Also, if you enabled session stickiness, ensure it's routing requests correctly (not dumping everyone onto a single overloaded container).

This is the most important step! When your colleague gets the "can't display page" error, immediately pull the Tomcat logs from the container. Use docker logs <your-tomcat-container-id> or check the AWS ECS task logs. Look for errors like:

• Too many open files (means your container has a file descriptor limit issue)
• ThreadPool exhausted (confirms your thread count is too low)
• Connection refused (points to network or connector problems)

These logs will tell you exactly what's breaking when multiple users connect.

While you said non-Docker works, it's worth double-checking if your web app has any hidden single-user locks. Some apps have test-mode settings that restrict sessions to one user—make sure that's not enabled in your containerized deployment.

Start with the logs, then work through the connector and resource settings—you'll likely find the culprit there. Good luck!

内容的提问来源于stack exchange，提问作者João Matos