Repository navigation
Leios: raise the endorser block ceiling to 600 kB - #2392
Merged
Merged
Conversation
Base automatically changed from
nfrisby/leios-issue-1071-validate-cert-asap
to
leios-prototype
October 9, 2026 20:56
nfrisby
approved these changes
Oct 9, 2026
nfrisby
left a comment
nfrisby
left a comment
Contributor
There was a problem hiding this comment.
Everything I commented on was about a comment being wrong; all should be easy to fix.
| -- parameters are first learned -- at node startup or on adopting the update | ||
| -- -- and not in the forge path. | ||
| -- | ||
| -- TODO Also check this when the parameter update is proposed, so that a |
Contributor
There was a problem hiding this comment.
Isn't that TODO the actual functionality of the era's guard rails script?
Contributor
Author
There was a problem hiding this comment.
Yeah, but nothing checks whether our code is consistent with the constitution / the guard rail script. So we can only check (in code) consistency with the pparams?
maxLeiosEbBytesSize goes from 512 KiB to 600,000 bytes, a base-10 value meant to equal the constitution's guardrail on maxEndorserBlockReferencesSize. guardLeiosWireLimit refuses to run with a parameter above it, so the guardrail can only follow a node release. 600 kB allows 16,216 references to transactions of 256 B or more (16,666 to smaller ones), so a 12 MB closure is reached once transactions average about 740 B. Below that the reference list binds, which is where simple payments sit. maxTxsPerEb becomes 17,142, which keeps worstCaseCacheTxCount (128 * 17,142 = 2.19M) at about half of the tx cache's 2^22 slots. The comments claiming the worst case overflows that table date from the 71,428 of the largeByteLimit derivation; they are corrected, and the txcache bench now runs at maxTxsPerEb rather than a stale 15,058. Every other mention of the 512 KiB figure in code and benches is updated.
ch1bo
force-pushed
the
ch1bo/leios-eb-ceiling-600kb
branch
from
October 9, 2026 21:51
fd1a560 to
d689925
Compare
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
maxLeiosEbBytesSizegoes from512 KiBto600,000 bytes(600KB), a base-10 value meant to equal the constitution's guardrail onmaxEndorserBlockReferencesSize.guardLeiosWireLimitrefuses to run with a parameter above it, so the guardrail can only follow a node release.See https://hackmd.io/ivfQtLgLQSOMn0QXYA3_rw?suggest-edit=5c2acebe-ee4a-479f-8825-ce637c458f25&utm_source=suggest-edit-card&utm_medium=icon for a working document about the guard rails.
600 kB allows 16,216 references to transactions of 256 B or more (16,666 to smaller ones), so a 12 MB closure is reached once transactions average about 740 B. Below that the reference list binds, which is where simple payments sit.
maxTxsPerEb becomes 17,142, which keeps worstCaseCacheTxCount
(128 * 17,142 = 2.19M) at about half of the tx cache's 2^22 slots. The comments claiming the worst case overflows that table date from the 71,428 of the largeByteLimit derivation; they are corrected, and the txcache bench now runs at maxTxsPerEb rather than a stale 15,058.