Updating the Readme file with perf optimizations and monitoring commands #240
Conversation
Performance optimizations and monitoring the monitor
Update README.md
| To see a real-time, dynamic view of these containers performance use `docker stats`. | ||
| After the initial launch you can improve the performance, by using some or all of the following options: | ||
| - consider running a local CL+EL. | ||
| - remove `--max-old-space-size=8192` from `start:prod` section of `ethereum-validators-monitoring/package.json` file. |
There was a problem hiding this comment.
| - remove `--max-old-space-size=8192` from `start:prod` section of `ethereum-validators-monitoring/package.json` file. |
There was a problem hiding this comment.
It doesn't bring new resources and even take them away. Removing this option can lead to a heap overflow in JS
There was a problem hiding this comment.
When I ran the app locally I faced some quirks with this parameter. If it was presented, the app didn't consume the whole memory available for the container for some reason. When I removed this param, the issue was gone.
You are correct that without this key the app can face heap overflow and we faced this issue in other environments, so, it is mandatory. I think we should remove the max-old-space-size key from the start:prod script and add the ability to pass this value to the app via the env variable. It will allow users to do more fine-grained configuration and also can resolve conflicts with limits in containers when the app is run via the docker-compose. But it should be done in a separate PR. As far as your note regarding the current PR, I agree, it's better not to include this suggestion in Readme as a performance optimization.
| resources: | ||
| limits: | ||
| cpus: '8' | ||
| memory: 16g |
There was a problem hiding this comment.
| memory: 16g |
Even at peak loads you don't need more than 3g, so I would suggest keep the current value (as in the current docker-compose) - 8g
| To tail the logs of your instantiation you can use `docker compose logs -f`. Watch out for connection errors and other issues especially regarding `ethereum-validators-monitoring` process. You can choose to ignore alerting related errors if you haven't configured this feature yet. | ||
| To see a real-time, dynamic view of these containers performance use `docker stats`. | ||
| After the initial launch you can improve the performance, by using some or all of the following options: | ||
| - consider running a local CL+EL. |
There was a problem hiding this comment.
Let's add a few words about:
- Increasing CPU on the host on which the application runs
- If running local nodes is not an option for you - reduce possible obstacles between remote nodes and application (firewalls, etc.) - anything that can slow down the connection
|
@astrostakers thanks for your suggestions! |
|
@astrostakers thank you so much for your help and attention to this project! |
Adding performance optimizations and monitoring the monitor sections to the readme file.