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
storage: node liveness records shouldn't use MVCC #37708
Comments
I agree that MVCC isn't useful here. My expectation is that we'll move away from node liveness (for example to @andy-kimball's proposed failure detector) and that it won't be worth it to do something in the interim (was this issue motivated by something, for example a stability problem? I know we've had some in the past, but thought we had successfully worked around them for now). |
No, it wasn't motivated by anything in particular.
I was actually wondering about that. I thought it wouldn't, but then I couldn't find what prevents it. Now I did. |
We have marked this issue as stale because it has been inactive for |
definitely still relevant |
We have marked this issue as stale because it has been inactive for |
These records currently use MVCC, but that seems ridiculous. We're wasting space and causing splits in a sensitive region of the key space.
I think we should move away from MVCC, but I'm not entirely sure about how the migration would work. @tbg have you ever thought about this?
I've also found this random note that wants the liveness record key prefix changed, which perhaps could be done at the same time.
#18276 (comment)
cc @nvanbenschoten
Jira issue: CRDB-4421
The text was updated successfully, but these errors were encountered: