feat(MR): [MR-649] Roll out best-effort calls to the first few subnets #3816
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Replace the on-off flag with an enum of 4 progressive rollout stages: * Stage 0: Trap on related API calls; and reject best-effort requests when routing (status quo). * Stage 1: Silently ignore `ic0_call_with_best_effort_response()` calls, falling back to guaranteed response calls; and reject best-effort requests when routing. * Stage 2: On system subnets, silently ignore `ic0_call_with_best_effort_response()` calls, falling back to guaranteed response calls; and reject best-effort requests that should be routed to system subnets. * Stage 3: Fully enable API calls; always route best-effort requests. Rollbacks (incremental or direct) are also supported, but it's probably a bad idea to roll back to stage 0, as any canisters already using best-effort calls would break. Badly.
Co-authored-by: stiegerc <[email protected]>
…e rolling out to specific subnets only.
derlerd-dfinity
approved these changes
Feb 6, 2025
Base automatically changed from
alin/MR-649-no-best-effort-calls-to-system-subnets
to
master
February 7, 2025 08:36
List a few subnets in the BestEffortResponsesFeature::SpecificSubnets under the FeatureFlags default. This will have the ic0_call_with_best_effort_response() API call on these subnets actually set a deadline on the calls being constructed. And allow best-effort requests to be routed between these subnets only. Also reduce the default subnet_message_memory_capacity on all subnets from 25 GB to 15 GB, to make place for a potential 5 GB of best-effort messages plus a potential 1 M callbacks. Apart from brief spikes, usually associated with heartbeat / global timer storms, subnet message memory usage is virtually always in the low single digit GBs, so the reduction of the limit will make no practical difference.
alin-at-dfinity
force-pushed
the
alin/MR-649-partially-roll-out-best-effort-calls
branch
from
2e80716 to
0fc745a
Compare
github-actions
bot
added
@execution
@consensus
@ic-message-routing-owners
@languages
labels
michael-weigelt
approved these changes
Feb 7, 2025
alin-at-dfinity
deleted the
alin/MR-649-partially-roll-out-best-effort-calls
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
List a few subnets in the BestEffortResponsesFeature::SpecificSubnets under the FeatureFlags default. This will have the ic0_call_with_best_effort_response() API call on these subnets actually set a deadline on the calls being constructed. And allow best-effort requests to be routed between these subnets only.
Also reduce the default subnet_message_memory_capacity on all subnets from 25 GB to 15 GB, to make place for a potential 5 GB of best-effort messages plus a potential 1 M callbacks. Apart from brief spikes, usually associated with heartbeat / global timer storms, subnet message memory usage is virtually always in the low single digit GBs, so the reduction of the limit will make no practical difference.