Galaxy containers are packaged environments that are optimized to run Meteor. Each container runs one copy of an app.
Each container has CPU and Memory allowance. Galaxy combines the pooled resources of all containers to track overall app performance. You can dive into individual container performance using the graphs on the app’s containers page.
Since every app has unique architecture and performance characteristics, adjusting the container size helps you meet the specific resource needs of your app.
- Compact: 512 MB/0.5 ECU
- Standard: 1 GB/1 ECU
- Double: 2 GB/2 ECU
- Quad: 4 GB/4 ECU
ECU is defined by Amazon and is designed to provide a relative measure of processing power. The abbreviation stands for EC2 Compute Unit.
View logs from a specific container
Accessing the logs from a specific container can help you diagnose any unexpected behavior. Click on the icon near the container name to see the logs from this container.
Connect to a specific container
If you suspect a container is misbehaving or would like to access container-specific information, you may want to connect to that specific container to learn more. From your app’s containers page, click on the icon near the container’s name to see your app as served by this container.
Since we re-route to a new container when your container is unavailable, you should verify which container subscriptions and remote method calls are being routed through.
Kill a container
In instances when application code behaves unexpectedly you may want to kill a container. This forcefully stops the running process and launches your Meteor app in a fresh container.
You can kill a container by going to the designated container on your app’s containers page and clicking the kill button below the title.
Containers may be restarted for several reasons, such as:
- A bug in an app caused memory to spike, which caused its container to be killed and then restarted.
- The app was working correctly, but it was overwhelmed by the number of connections or the load placed on it. Consequently, its container ran out of memory, was killed and restarted.
- Our system underwent an upgrade, and as part of the upgrade, all containers were restarted.
If constant uptime is a core requirement, we recommend running your app on more than 1 container. Since periodic system updates and container restarts can be expected on Galaxy, this can help prevent unwanted downtime. When an app runs on multiple containers, traffic can be routed elsewhere when a single container becomes unresponsive.
Please note that, if you do choose to run a 1 container app, downtime will be unavoidable in the event of a container restart or a new deployment. Due to the need for system maintenance, we cannot guarantee that containers will not be restarted.