Graceful shutdown is ignored by Tilt about tilt

[nicks]

thanks for the report! But i was unable to reproduce this problem, given the steps provided.

Here's what I tried:

• i modified the example project to use graceful termination - tilt-dev/tilt-example-html@1bf5c3b
• ran tilt up
• triggered a reload of the resource
• used stern example-html to confirm that the old pod received a TERM signal, then stayed up for a while until kubernetes killed it.

what makes you think it's not respecting graceful termination?

[nicks]

side note -- under the hood, tilt just calls kubectl delete, and lets kubernetes handle shutdown, tilt doesn't send signals to the process directly -- so i'm not sure even how this bug could happen

[o-shevchenko]

Thanks for the answer and details @nicks !
What I see is that pod is killed right after trigger upgrade. Let me dig dipper and check if I see the same behavior for kubectl delete for pods created by Tilt. However, helm upgrade for the same Helm charts works fine, and pods are waiting for graceful termination.