Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Provide subcommand for scaling out a ring #174

Open
a1exsh opened this issue Aug 30, 2017 · 5 comments
Open

Provide subcommand for scaling out a ring #174

a1exsh opened this issue Aug 30, 2017 · 5 comments

Comments

@a1exsh
Copy link
Collaborator

a1exsh commented Aug 30, 2017

Since introduction of AUTO_BOOTSTRAP parameter, it is no longer safe to blindly follow the documented "Scale out" procedure. We should have a dedicated subcommand to add a node to the ring.

a1exsh added a commit that referenced this issue Aug 30, 2017
Temp. fix to address #174, needs a proper automation.
@lmineiro
Copy link
Contributor

lmineiro commented Sep 8, 2017

Why do you say it's not safe to extend the ring without automatic bootstrap?

@a1exsh
Copy link
Collaborator Author

a1exsh commented Sep 8, 2017

I believe the new node will claim the tokens and start serving reads immediately, but it doesn't have the data. Do you think it's not the case?

@lmineiro
Copy link
Contributor

lmineiro commented Sep 8, 2017

I believe you're right. I still see a couple of good use cases where this is The Right Thing To Do: 2 AZs to 3 AZs, for ex, to avoid having one first super node.
Apart from read consistency level of ALL (which automatically repairs) I'd see only the option of disabling gossip on the node (marking it as down) so that it wouldn't serve reads.
Maybe we should test once again if a new node, that wasn't automatically bootstrapped, starts serving reads and, if it indeed it still does, if the option of disabling gossip works. CL = ALL would be overkill

@a1exsh
Copy link
Collaborator Author

a1exsh commented Sep 8, 2017

Interesting thoughts. In any case someone who is not aware of these details could impact the cluster severely, if the exiting doc is followed as is.

@lmineiro
Copy link
Contributor

lmineiro commented Sep 8, 2017

👍

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants