feat: Add startup_script_timeout and delay_login_until_ready
#84
Conversation
| Type: schema.TypeString, | ||
| ForceNew: true, | ||
| Optional: true, | ||
| Description: "A script to run before the agent is stopped. The script should exit when it is done to signal that the workspace can be stopped.", |
There was a problem hiding this comment.
Updated description here, I considered commenting it out or adding (NOT YET IMPLEMENTED.), but we'll likely do this soon after we get the startup script changes in.
docs/resources/agent.md
Outdated
| @@ -46,13 +46,15 @@ resource "kubernetes_pod" "dev" { | |||
|
|
|||
| ### Optional | |||
|
|
|||
| - `allow_login_before_ready` (Boolean) Allow users to login to the workspace before the agent is ready. Note that, when enabled, the agent could still be executing the startup script and the workspace in an incomplete state when logging in. | |||
There was a problem hiding this comment.
@kylecarbs does this match what you were suggesting in coder/coder#5749 (comment)?
I interpreted it as an option to allow/prevent it. Is that how you intended it? Or did you intend it to be the "default behavior for login before ready"? If it's the latter, we could rename this to be just login_before_ready instead of allow.
There was a problem hiding this comment.
I might invert the naming scheme because, eventually we'll make this default to blocking login before ready, and ideally it'd be false by default.
Apart from that we're aligned!
There was a problem hiding this comment.
I was going a back and forth on the naming a bit:
wait_until_ready
stop_login_before_ready
delay_login_until_ready
block_login_until_ready
prevent_login_before_ready
disallow_login_before_ready
deny_login_before_ready
I think delay_login_until_ready is most descriptive and flexible option. This would allow us to implement --skip-delay-login --confirm-i-really-want-to-log-in-to-my-partially-started-workspace flag for users trying to SSH in. 😄
Likewise if admins want to truly block the behavior (i.e. startup script must complete before you can login), we can add a second boolean to actually prevent the behavior.
Alternatively we could have something like login_before_ready: delay, login_before_ready: no_skip, login_before_ready: yes.
Edit: 163bba0
Re: RFC, is there any specific questions you have after reading coder/coder#5749? I felt it adequately describes what will be implemented but I'm happy to answer any questions. IMO it won't be a very complex change since it's focused towards fixing the |
startup_script_timeout and allow_login_before_readystartup_script_timeout and delay_login_until_ready
|
Yes, I have a feeling that we're missing a short paragraph mainly describing changes: It's hard to figure out from |
Ref: coder/coder#5749