Skip to content

Bench/leios new - #6622

Draft
fmaste wants to merge 260 commits into
masterfrom
bench/leios-new
Draft

fmaste wants to merge 260 commits into
masterfrom
bench/leios-new

Conversation

@fmaste

@fmaste fmaste commented Jul 17, 2026

Copy link
Copy Markdown
Contributor

Description

Add your description here, if it fixes a particular issue please provide a
link
to the issue.

Checklist

  • Commit sequence broadly makes sense and commits have useful messages
  • New tests are added if needed and existing tests are updated. These may include:
    • golden tests
    • property tests
    • roundtrip tests
    • integration tests
      See Running tests for more details
  • Any changes are noted in the CHANGELOG.md for affected package
  • The version bounds in .cabal files are updated
  • CI passes. See note on CI. The following CI checks are required:
    • Code is linted with hlint. See .github/workflows/check-hlint.yml to get the hlint version
    • Code is formatted with stylish-haskell. See .github/workflows/stylish-haskell.yml to get the stylish-haskell version
    • Code builds on Linux, MacOS and Windows for ghc-9.6 and ghc-9.12
  • Self-reviewed the diff

Note on CI

If your PR is from a fork, the necessary CI jobs won't trigger automatically for security reasons.
You will need to get someone with write privileges. Please contact IOG node developers to do this
for you.

crocodile-dentist and others added 30 commits July 18, 2025 12:50
…11 Demo

NOTE:
- Tests and benchmarks fail to build, and that's ok
More consistent traces between immdb-server and cardano-node
ch1bo and others added 27 commits September 7, 2026 01:03
Expose LeiosDb volatile/immutable partition sizes as node metrics
The SQLite LeiosDB starts a writer worker holding the only write
connections plus maintenance threads, and until now nothing ever stopped
them. Open it through consensus' openLeiosDBSQLite and run its teardown
when the node stops: everything else has wound down by then, so this
flushes the pending writes, stops the threads and closes the connections.

Depends on the writer-lifecycle work on ouroboros-consensus #2298.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Both partitions resolved against the non-immutable path, so a node
configured with MultipleDbPaths put the LeiosDB's immutable partition on
the performant volume -- the one that option exists to keep small.

The partitions split the same way the node's own do: the volatile one
churns and is swept, the immutable one only grows. Resolve each against
the matching path. No change under OnePathForAllDbs, where both are the
same directory.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A node could not start on a fresh database directory. The ChainDB
refuses a directory that holds files but no marker of its own, and
opening the LeiosDB creates its SQLite files in that same directory
whenever the node runs on a single database path -- which happens
before 'Node.run' gets as far as writing the marker.

Run that check here instead, before the LeiosDB is opened. It is the
same 'checkDbMarker' the node makes later and it is idempotent: with the
marker present it only verifies the network magic. Nothing else changes
order, and no lazy construction is needed to get it.

Also drops the leiosDbCopyQueueFull counter, whose trace constructor is
gone with the copy queue.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Matches the tools: leios.vol.db and leios.imm.db.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The SQLite backend took a path per partition. Now that each partition
follows the node's own DatabasePath -- volatile beside volatile/,
immutable beside immutable/ -- a second way to name them can only
contradict it: an absolute pair could put the growing partition on the
performant volume, or put both back inside a directory the ChainDB
expects to own. So the backend is all the option carries now, and the
paths are derived.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The DB creates its own partition directories now, so the node does not.
All keys reach the Leios voting logic, which votes once per committee
seat any of them holds: a rotation pair keeps voting across the epoch
boundary, a bundle of pool keys votes for every seat it covers.
Leios: Accept a JSON array of BLS key envelopes in --shelley-bls-key
Update the counter state, otherwise we will never have more than one sample.
…/2026w39

leios-prototype-2026w29 integrations
@github-actions

Copy link
Copy Markdown

This PR is stale because it has been open 45 days with no activity.

@github-actions github-actions Bot added the Stale label Nov 11, 2026

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.