-
Notifications
You must be signed in to change notification settings - Fork 217
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
Add basic backpressure mechanism if primary has too many in-flight transactions #5692
Add basic backpressure mechanism if primary has too many in-flight transactions #5692
Conversation
Report error rate, fail if non-zero Pros: safe, provided an exception is made for Report error rate, don't fail if non-zero Pros: can run tests close to the cap in CI Report error rates and plot them Pros: can run tests close to the cap in CI |
committable_threshold_backpressure@76588 aka 20230928.29 vs main ewma over 20 builds from 76284 to 76583 Click to see tablemain
committable_threshold_backpressure
|
Summary of discussion with @eddyashton:
|
Co-authored-by: Eddy Ashton <[email protected]>
💔 All backports failed
Manual backportTo create the backport manually run:
Questions ?Please refer to the Backport tool documentation and see the Github Action logs for details |
💔 All backports failed
Manual backportTo create the backport manually run:
Questions ?Please refer to the Backport tool documentation |
…ansactions (microsoft#5692) (cherry picked from commit d1d9d8e) # Conflicts: # CHANGELOG.md # tests/infra/basicperf.py # tests/infra/remote.py
Resolve #3871 and #5690.
Added a
consensus.max_uncommitted_tx_count
, which caps the maximum number of uncommitted transactions a primary allows before pushing back with HTTP errors. An exception is carved out for the/node
frontend, to avoid hampering diagnostic attempts on a primary that may be stuck at the cap (for example because >f
nodes have died).Working on the tests (see comments), but the change itself is ready to review.