-
Notifications
You must be signed in to change notification settings - Fork 2.1k
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
Receive: Dont rebatch already replicated requests #5818
Conversation
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.
👍
Hopefully in the long term we can end up with your idea of calculating all replicas for all batches in one place, but this should be good enough for now. |
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.
LGTM :)
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.
LGTM!
Signed-off-by: Matej Gera <matejgera@gmail.com>
Signed-off-by: Matej Gera <matejgera@gmail.com>
Signed-off-by: Matej Gera <matejgera@gmail.com>
ea3214c
48faecd
to
ea3214c
Compare
Can this be rebased? 🙂 |
Closing as this has been superseded by #5910 |
Changes
Part of #5807.
Since change in #5604, we are already hashing each time series by endpoint and replica, meaning that replicated requests will now always be sent to the actual endpoint to which they are supposed to be written. This means the 're-batching' of series upon receiving a replicated request is redundant and can be skipped.
Verification
Tests are succeeding locally. I also added one more benchmark with replicated request to see changes:
Note
We should make obvious to our users that this change is not fully compatible between older and newer versions of receive. In particular if any user would be running a mix of pre-
0.28.0
receivers and receivers with this change (which can happen during a rollout), the batching would not work correctly. It is therefore important to instruct users to first update their receivers to version to either0.28.0
or0.29.0
and only then to newer version with this change.