You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Apr 18, 2024. It is now read-only.
I've deployed crucible on docker and I notice that if I start 2 (lengthy) test runs at the same time, one test run will not begin until the first one completes.
I don't see this problem on the public site: it this because you are deploying multiple instances of either the web or task containers behind a load balancer?
Or is this unexpected behaviour pointing to some configuration or other issue with my deployment?
The text was updated successfully, but these errors were encountered:
Hi @pcosmog -- our production instance doesn't use docker. It is on a single VM, but it starts three task runners that execute the actual tests. Our docker configuration only starts a single test runner, and therefore can only run one set of tests at once.
I put up a branch that should increase the number of task runners that docker runs: #309
Could you pull down that branch and see if that configuration better suits your need?
One other note, if you are using this in a long-lived production instance, we also have a cron job that watches for tasks that have stalled. For very large test runs that are hitting servers with a lot of data, we've found our task runners can be a little unreliable. So that watcher job looks out for stalled test jobs and resumes them in a new worker. That isn't currently configured to run in our docker container, but we can look to add that there as well if you are expecting to use this in a non-development environment.
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
I've deployed crucible on docker and I notice that if I start 2 (lengthy) test runs at the same time, one test run will not begin until the first one completes.
I don't see this problem on the public site: it this because you are deploying multiple instances of either the web or task containers behind a load balancer?
Or is this unexpected behaviour pointing to some configuration or other issue with my deployment?
The text was updated successfully, but these errors were encountered: