op-node: Handle duplicate outgoing supervisor events

[sebastianst]

Description

• Removes emitting StepReqEvent inside SyncDeriver.onIncomingP2PBlock - this was causing an avalanche of exhaust l1 events
• Deduplicate sending the same supervisor events from a managed node with a TTL until the same message is resend.
  • Complete duplication is also dangerous, as a supervisor may be restarting and miss state.

Tests

Added unit test of new eventTimestamp type and unit test of event handling behavior of ManagedNode.

Additional context

As long as we haven't fully verified that the node isn't sending a burst of duplicate events in some edge cases, it is safer to deduplicate sending events to the supervisor in this way.

Metadata

Fixes #16284

[axelKingsley]

Wondering out loud: would this timestamp TTL also be useful for reducing things like UpdateCrossUnsafeEvent during deep sync operations?

[codecov]

Codecov Report

Attention: Patch coverage is 75.32468% with 19 lines in your changes missing coverage. Please review.

Project coverage is 45.99%. Comparing base (7adb589) to head (3bf2116).
Report is 22 commits behind head on develop.

Files with missing lines	Patch %	Lines
op-node/rollup/interop/managed/system.go	69.35%	17 Missing and 2 partials ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop   #16441      +/-   ##
===========================================
- Coverage    46.91%   45.99%   -0.92%     
===========================================
  Files         1394     1340      -54     
  Lines       112800   108561    -4239     
===========================================
- Hits         52915    49938    -2977     
+ Misses       56150    55028    -1122     
+ Partials      3735     3595     -140     

Flag	Coverage Δ
cannon-go-tests-64	?
contracts-bedrock-tests	96.24% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
op-node/rollup/driver/state.go	0.00% <ø> (ø)
op-node/rollup/interop/managed/event_timestamp.go	100.00% <100.00%> (ø)
op-node/rollup/interop/managed/system.go	28.80% <69.35%> (+15.68%)	⬆️

... and 71 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[sebastianst]

Wondering out loud: would this timestamp TTL also be useful for reducing things like UpdateCrossUnsafeEvent during deep sync operations?

@axelKingsley yes we could totally leverage the new eventTimestamp type and accompanying pattern to deduplicate a few other events hammering the system. Good idea for a follow-up.

edit: although the cleaner fix imo is to make the code more robust and not send duplicates in the first place. I would like for us to switch some events/triggers to be range-based instead of single-block based.

[protolambda]

The changes make sense, although I am still slightly concerned with the L1 traversal itself.

[sebastianst]

The changes make sense, although I am still slightly concerned with the L1 traversal itself.

My initial implementation had a bug where the derived event was deduplicated based solely on the L1 block ID, which of course doesn't make sense for the derived event, which is also emitted for every new L2 block. This event is now deduplicated based on a combined identifier of the L1 derivedFrom and L2 derived block id. Not sure if this is related to your concerns. Edit: I could actually revert that change. It's enough to deduplicate based on L1 ref for the L1 traversal trigger and L2 ref for the local unsafe trigger.

[teddyknox]

Stamping so that we can optimistically merge and trigger post-merge CI.