-
Notifications
You must be signed in to change notification settings - Fork 3.8k
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
release-19.1: storage: remove dependency from replication queue to the split queue #38588
release-19.1: storage: remove dependency from replication queue to the split queue #38588
Conversation
Wanna give this a nightly or two?
|
I don't think it's particularly scary, but sure.
I've given a custom binary to the affected user anyway.
…On Mon, Jul 1, 2019 at 2:35 PM Tobias Grieger ***@***.***> wrote:
Wanna give this a nightly or two?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#38588?email_source=notifications&email_token=AAC4C4PHDTSGKL4DT4GXETLP5JE6XA5CNFSM4H4U36KKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODY67HPY#issuecomment-507376575>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAC4C4M6725IVRDTG6WPH3TP5JE6XANCNFSM4H4U36KA>
.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
a night or two have passed :P
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @nvanbenschoten)
LGTM |
bors r+ |
Build failed |
7983c3d
to
6109109
Compare
Before this patch, the replication queue would refuse to process ranges that it thought needed splits because of their size. The idea was to let the split queue process them first. There's no reason for this, particularly since we've introduced the write-backpressuring mechanism which keeps ranges from getting too large (although not all requests are backpressured, so theoretically ranges could still grow larger than we'd like). The check had not considered unsplittable ranges (i.e. ranges with only (versions of) one row or some system ranges). These can't be split but the replication queue was disconsidering them none the less. This change was motivated by the unsplittable SystemConfigSpan range. This kind of system range is not backpressured, but we still don't want the replication queue to dismiss it when it's too large (since, again, it can't be split). Release note (bug fix): Ranges consisting of only one row (and historical versions of that row) are now correctly up-replicating.
6109109
to
0cfe6b1
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've removed the test from this backport. It doesn't work as written because the way to change default zone configs was different before 19.2. I don't think it's worth fixing.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @nvanbenschoten)
Backport 1/1 commits from #38529.
/cc @cockroachdb/release
Before this patch, the replication queue would refuse to process ranges
that it thought needed splits because of their size. The idea was to let
the split queue process them first. There's no reason for this,
particularly since we've introduced the write-backpressuring mechanism
which keeps ranges from getting too large.
The check had not considered unsplittable ranges (i.e. ranges with only
(versions of) one row or some system ranges). These can't be split but
the replication queue was disconsidering them none the less.
Release note (bug fix): Ranges consisting of only one row (and
historical versions of that row) are now correctly up-replicating.