a year ago
I added the RAILWAYDEPLOYMENTDRAINING_SECONDS environment variable to my project and set it to 300 to give my old service 5 mins to finish whatever tasks it was doing, but now it's stuck in the REMOVING state and it stopped sending signal SIGTERM to the container. Will it ever send the SIGKILL signal or is there a manual way of stopping the container? Thanks!
3 Replies
FWIW, I never got my containers to be completely removed even after removing my graceful shutdown code. All subsequent deployments would get stuck in the REMOVING state.
Also, I don't know if this is related but I was testing out attaching/removing volumes while this was happening and it never fully detached the volume. It got stuck in an applying state and kept taking down my deployment. I expected the service to redeploy to unmount the volume but completely taking it down and deleting the running deployment seems bizarre. I didn't see anything in the docs about unmounting volumes stopping services, but maybe I missed it?
I ended up having to spin up a completely new project and re-deploying my services there so that project ID is probably no longer valid. Happy to explain further if it would be helpful to the team!
a year ago
hello!
do you think you would be able to provide steps to reproduce the issues mentioned? theres unfortunately nothing for us to look into with a deleted project.