graph/db: cross-version graph Store queries#10714
graph/db: cross-version graph Store queries#10714ellemouton wants to merge 6 commits intolightningnetwork:masterfrom
Conversation
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request refactors the graph database layer to support cross-version gossip data. By introducing materialized mapping tables, the system can now automatically select the preferred version of nodes and channels based on defined priority rules. This change simplifies the API for callers and ensures that graph traversal methods consistently return the most relevant data, regardless of whether it was announced via gossip v1 or v2. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request refactors the graph Store interface to support cross-version operations, enabling node and channel traversals to work across gossip v1 and v2. It introduces preferred mapping tables in the SQL backend and PreferHighest fetch helpers to prioritize v2 data while falling back to v1. Feedback was provided to implement a feature fallback in the no-cache path for directed channel traversals, ensuring consistency with the FetchNodeFeatures logic by providing the best available feature set when a specific gossip version lacks data.
| } | ||
| } | ||
|
|
||
| toNodeCallback := func() route.Vertex { |
There was a problem hiding this comment.
In the no-cache path, we should implement a fallback for node features similar to the logic in ChannelGraph.FetchNodeFeatures. This ensures that if a node has v1 features but no v2 features, we still provide the v1 features for pathfinding even if the channel announcement is v2. This maintains consistency between the cache-enabled and cache-disabled code paths.
// Try v2 first, fall back to v1 if the v2 features are empty. This
// ensures consistency with FetchNodeFeatures and provides the best
// available feature set for the node.
preferredFeatures := featuresByVersion[gossipV2]
if preferredFeatures.IsEmpty() {
preferredFeatures = featuresByVersion[gossipV1]
}
toNodeCallback := func() route.Vertex {There was a problem hiding this comment.
I think the current behaviour is actually more correct here. Tying ToNodeFeatures to the channel's gossip version ensures the features match the protocol version of the channel advertisement. A v2 preferred channel should use v2 features, and a v1 preferred channel should use v1 features.
Using a single "best" feature set across all channels would mean a v1 channel could get v2 features assigned, which is a version mismatch.
The underlying ambiguity (empty features vs absent node for a given version) is a known limitation — the proper fix is to distinguish those two cases at the FetchNodeFeatures layer rather than papering over it here.
| cachedInPolicy.ToNodeFeatures = | ||
| featuresByVersion[lnwire.GossipVersion( | ||
| row.GraphChannel.Version, | ||
| )] |
There was a problem hiding this comment.
Use the pre-calculated preferredFeatures here to ensure the best available feature set is used for the node, regardless of the channel's gossip version.
| cachedInPolicy.ToNodeFeatures = | |
| featuresByVersion[lnwire.GossipVersion( | |
| row.GraphChannel.Version, | |
| )] | |
| cachedInPolicy.ToNodeFeatures = preferredFeatures |
ellemouton
force-pushed
the
g175-graph-db-cross-version-v2
branch
from
32f201a to
652f923
Compare
ellemouton
force-pushed
the
g175-graph-db-cross-version-v2
branch
from
652f923 to
24c9c8d
Compare
| WHERE (n1.pub_key = @pub_key OR n2.pub_key = @pub_key) | ||
| AND pc.scid > @scid | ||
| ORDER BY pc.scid | ||
| LIMIT @page_limit; |
There was a problem hiding this comment.
nit: it would be good to stay consistent with either the $ or @ syntax
| ORDER BY c.id | ||
| LIMIT $3; | ||
|
|
||
| -- name: ListPreferredChannelsWithPoliciesPaginated :many |
There was a problem hiding this comment.
naming-nit: we could make the naming more consistent
ListPreferredChannelsPaginated and ListPreferredNodeChannelsPaginated
or with the WithPolicies addition.
There was a problem hiding this comment.
perhaps it makes sense to shorten:
prunedNodes := make([]route.Vertex, len(nodeKeys))
for i, nodeKey := range nodeKeys {
err = db.UpsertPreferredNode(ctx, nodeKey)
if err != nil {
return nil, fmt.Errorf("recalc preferred node: %w", err)
}
pub, err := route.NewVertexFromBytes(nodeKey)
if err != nil {
return nil, fmt.Errorf("unable to parse pubkey from bytes: %w", err)
}
prunedNodes[i] = pub
}
return prunedNodes, nilsince the upsert is idempotent? but maybe an additional dedup can make sense
| } | ||
| } | ||
|
|
||
| // IsPublicNode is a helper method that determines whether the node with the |
| pageIDs []channelBatchIDs) (*batchChannelData, error) { | ||
|
|
||
| for i, ids := range allIDs { | ||
| channelIDs := make([]int64, len(pageIDs)) | ||
| policyIDs := make([]int64, 0, len(pageIDs)*2) | ||
| for i, ids := range pageIDs { | ||
| channelIDs[i] = ids.channelID |
There was a problem hiding this comment.
nit: change maybe not needed
| ctx, models.NewShellNode(lnwire.GossipVersion2, shellPub), | ||
| )) | ||
|
|
||
| nodesByPub := make(map[route.Vertex]*models.Node) |
There was a problem hiding this comment.
tiny nit: the map de-duplicates the results and would let a duplicate node pass, same with channels below
| -- name: DeletePreferredNode :exec | ||
| DELETE FROM graph_preferred_nodes WHERE pub_key = $1; |
There was a problem hiding this comment.
looks like the delete queries are unused as we rely on the cascade functionality
| GetChannelAndNodesBySCID(ctx context.Context, arg sqlc.GetChannelAndNodesBySCIDParams) (sqlc.GetChannelAndNodesBySCIDRow, error) | ||
| HighestSCID(ctx context.Context, version int16) ([]byte, error) | ||
| ListChannelsByNodeID(ctx context.Context, arg sqlc.ListChannelsByNodeIDParams) ([]sqlc.ListChannelsByNodeIDRow, error) | ||
| ListPreferredDirectedChannelsPaginated(ctx context.Context, arg sqlc.ListPreferredDirectedChannelsPaginatedParams) ([]sqlc.ListPreferredDirectedChannelsPaginatedRow, error) |
There was a problem hiding this comment.
ListNodesPaginated seems to be unused
Some other ones that could need a check:
GetNodesByIDsDeleteNodeGetZombieChannelsSCIDsGetPruneHashByHeightGetPruneEntriesForHeightsGetClosedChannelsSCIDsListChannelsWithPoliciesPaginatedGetChannelsByIDs
Add two precomputed mapping tables that track the "best" gossip version for each unique node (pub_key) and channel (SCID): - graph_preferred_nodes: pub_key -> node_id - graph_preferred_channels: scid -> channel_id Priority for nodes: v2 announced > v1 announced > v2 shell > v1 shell. Priority for channels: v2 with policies > v1 with policies > v2 > v1. These tables enable simple indexed-join queries for cross-version traversal (ForEachNode, ForEachChannel, ForEachNodeDirectedChannel) without expensive per-row COALESCE subqueries. The tables are populated from existing data during the migration and maintained by upsert/delete queries on every write path (added in the next commit).
Add four new version-agnostic Store methods that return the highest available gossip version for a channel: - FetchChannelEdgesByIDPreferHighest - FetchChannelEdgesByOutpointPreferHighest - GetVersionsBySCID - GetVersionsByOutpoint The KV store delegates to v1; the SQL store iterates versions from highest to lowest, returning the first with policies (falling back to the highest version without).
Add UpsertPreferredNode and UpsertPreferredChannel calls to every Store write path so the preferred mapping tables stay consistent: - upsertSourceNode, upsertNode, maybeCreateShellNode, DeleteNode - insertChannel, updateChanEdgePolicy, DeleteChannelEdges - pruneGraphNodes CASCADE deletes on the underlying graph_nodes and graph_channels tables automatically clean up preferred entries when a version is removed.
Drop the GossipVersion parameter from ForEachNode and ForEachChannel. Both methods now iterate across all gossip versions, yielding one result per unique pub_key or SCID using the preferred mapping tables for pagination. Wire ChannelGraph.FetchChannelEdgesByID and FetchChannelEdgesByOutpoint through the PreferHighest variants. Add GetVersionsBySCID and GetVersionsByOutpoint wrappers to ChannelGraph.
Update ForEachNodeDirectedChannel and FetchNodeFeatures to work across gossip versions. ForEachNodeDirectedChannel now checks for a preferHighestNodeDirectedChanneler interface on the store (implemented by SQLStore) to stream cross-version directed channels in a single paginated query against the preferred-channel mapping table. The fallback path iterates v2 then v1, buffering results to deduplicate by SCID. FetchNodeFeatures tries v2 first, falling back to v1 when v2 features are empty. The cache population comment is updated to note that v1-then-v2 iteration order naturally gives v2 precedence via key collision overwrite.
ellemouton
force-pushed
the
g175-graph-db-cross-version-v2
branch
from
24c9c8d to
b66429f
Compare
github-actions
Bot
added
the
severity-critical
PR Severity: CRITICAL
Critical (6 files):
High (4 files):
Low (1 file):
AnalysisThis PR introduces SQL migration 15 (000015_graph_preferred_lookups) adding new indexes/tables to the graph SQL schema, with corresponding sqlc-generated code and updated graph store implementations. Database migrations are always CRITICAL: they are irreversible in production and affect data integrity. Key concerns for reviewers:
To override, add a severity-override-{critical,high,medium,low} label. |
|
thanks @bitromortac - addressed 👍 |
|
@bhandras: review reminder |
4 participants
Summary
This PR makes the graph
Storeinterface cross-version so thatForEachNode,ForEachChannel, andForEachNodeDirectedChannelwork across gossip v1 and v2, returning one preferred entry per pub_key/SCID. It also addsPreferHighestfetch helpers andGetVersionsqueries so callers can retrieve channels without knowing which gossip version announced them.Part of the gossip v2 epic: #10293
Spun off from: #10656
Design
Two new materialised mapping tables are introduced:
graph_preferred_nodes— one row per uniquepub_key, pointing at the bestgraph_nodesrow. Priority: v2 announced > v1 announced > v2 shell > v1 shell.graph_preferred_channels— one row per unique SCID, pointing at the bestgraph_channelsrow. Priority: v2 with policies > v1 with policies > v2 bare > v1 bare.These tables are maintained on every write path (
AddNode,AddChannelEdge,UpdateEdgePolicy,DeleteNode,DeleteChannelEdges,PruneGraphNodes) viaUpsertPreferredNode/UpsertPreferredChannelSQL queries, and seeded by a migration for existing data.Changes by commit
sqldb: add preferred-node and preferred-channel mapping tables— migration 000014 creating the tables and populating them from existing data.graph/db: add PreferHighest fetch methods and GetVersions queries— newStoreinterface methods:FetchChannelEdgesByIDPreferHighest,FetchChannelEdgesByOutpointPreferHighest,GetVersionsBySCID,GetVersionsByOutpoint. KV implementations delegate to v1.graph/db: wire preferred-table maintenance into write paths— callsUpsertPreferredNode/UpsertPreferredChannelon all SQL write paths.graph/db: make ForEachNode and ForEachChannel cross-version— removes theGossipVersionparameter fromForEachNodeandForEachChannelon theStoreinterface, replacing the underlying queries with preferred-table-backed paginated queries.graph/db: implement cross-version node traversal— addsForEachNodeDirectedChannelPreferHighestfor cache-disabled SQL path, updatesFetchNodeFeaturesto prefer v2 features with v1 fallback.docs: add release notePerformance
Benchmarked on Apple M1 Pro against a mainnet v1-only graph (16,216 nodes, 51,239 channels). No regressions observed — the preferred-table JOIN adds negligible overhead:
Test plan
TestPreferHighestAndGetVersions,TestPreferHighestForEachNode,TestPreferHighestForEachChannel,TestPreferHighestNodeTraversal,TestPreferHighestNodeDirectedChannelTraversal,TestDeleteNodePreferredRecomputationisSQLDBguard