fix(grpc): return actual earliest_store_height in node.Status endpoint

[Cordtus]

Description

Closes: #15463

Add EarliestVersion() to the CommitMultiStore interface to enable querying the earliest available state height.
This addresses the longstanding TODO in the Status gRPC endpoint that was returning 0 for EarliestStoreHeight.

Problem

Indexers and tooling need to know which heights a pruned node can serve queries for. Currently:

• earliest_block_height in CometBFT's STATUS RPC returns the earliest block, not the earliest state
• The EarliestStoreHeight field in cosmos.base.node.v1beta1.Status was hardcoded to 0

Solution

• Add EarliestVersion() int64 to the CommitMultiStore interface (not MultiStore, since CacheMultiStore lacks root DB access)
• Implement persistent tracking in rootmulti.Store:
• Stores earliest version at s/earliest key
• Defaults to 1 for unpruned chains
• Updates after successful pruning in PruneStores()
• Pass CommitMultiStore to the node gRPC service to populate EarliestStoreHeight

Changes

File	Change
store/types/store.go	Add EarliestVersion() to `
client/grpc/node/service.go	Accept CommitMultiStore, use EarliestVersion() in Status response
store/rootmulti/store.go	Implement EarliestVersion(), GetEarliestVersion(), flushEarliestVersion(), update PruneStores()
store/rootmulti/store_test.go	Add 3 unit tests
runtime/app.go	Pass CommitMultiStore to service registration
simapp/app.go	Pass CommitMultiStore to service registration

Testing

• TestEarliestVersion - basic functionality, default value
• TestEarliestVersionWithPruning - tracks updates after pruning
• TestEarliestVersionPersistence - persists across restarts
• System test: TestNodeStatus - gRPC endpoint returns valid heights

API Changes

Breaking change for any code implementing CommitMultiStore — must now implement EarliestVersion() int64.
No impact on MultiStore or CacheMultiStore implementations.

[fmorency]

Would love to get this backported to 0.50.x. It would enable fixing manifest-network/yaci#28

[aljo242]

This seems like a nice feature but will need more extensive testing such as using systemtest if we are going to include

[Cordtus]

Added systemtest for the node.Status gRPC endpoint:

• Basic test confirms valid heights are returned
• Pruning test configures aggressive state pruning and confirms earliest_store_height increases afterward

Also updated the service to use CommitMultiStore directly since query context provides a CacheMultiStore which doesn't support version queries.

Run locally:

make build && cp build/simd tests/systemtests/binaries/
cd tests/systemtests && go test -tags system_test -v -run TestNodeStatus

[fmorency]

@aljo242 is there anything blocking this PR?

[Cordtus]

Bump. Would like to be sure there is nothing else needed from my end here.

[Cordtus]

@aljo242 is there anything I can do to help this one along?

[aljo242]

@technicallyty to review this

[codecov]

Codecov Report

❌ Patch coverage is 72.22222% with 10 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.70%. Comparing base (0deda3b) to head (3a3c034).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
client/grpc/node/service.go	44.44%	5 Missing ⚠️
store/rootmulti/store.go	80.76%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #25647      +/-   ##
==========================================
+ Coverage   68.23%   70.70%   +2.46%     
==========================================
  Files         894      895       +1     
  Lines       58150    58217      +67     
==========================================
+ Hits        39680    41161    +1481     
+ Misses      18470    17056    -1414     

Files with missing lines	Coverage Δ
runtime/app.go	65.34% <100.00%> (ø)
store/types/store.go	75.00% <ø> (+45.00%)	⬆️
client/grpc/node/service.go	62.96% <44.44%> (-1.04%)	⬇️
store/rootmulti/store.go	82.93% <80.76%> (+34.17%)	⬆️

... and 33 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[technicallyty]

This seems like a nice feature but will need more extensive testing such as using systemtest if we are going to include

i'm fine with merging this with systemtests, but i would like to use systemtest more sparingly in the future. systest runs a local 4 node network, which is not necessary for queries like this. an e2e integration test is more than enough for these types of PRs

[technicallyty]

is there a reason why we can't use the sdkCtx.MultiStore() to achieve this functionality?

[Cordtus]

Not that I know -- I can change it if that seems like a better way to go.

Edit: we would have to add a ctx.EarliestHeight() to it first, which is more consistent but I guess more code as well. Which do you think is the better path?

[technicallyty]

can you explain why it would have to be a method on ctx?

[Cordtus]

It doesn't — my bad. I moved EarliestVersion() from MultiStore to CommitMultiStore instead.

The query context's MultiStore() is an in-memory CacheMultiStore — created in CreateQueryContext via CacheMultiStoreWithVersion(height) with no reference to the root DB or its metadata.

Placing it on CommitMultiStore means CacheMultiStore doesn't need to implement it at all and avoids a stub that just panics (like LatestVersion() currently does on CacheMultiStore, which is the bad example I was following originally). This way the query/service receives CommitMultiStore at registration.

[technicallyty]

ok so if we go that route, we could just get EarliestHeight from ctx and then we don't have to pass around the cms? i'd prefer that

[Cordtus]

I edited the original description and cleaned it up a bit. This makes more sense for sure. Sorry for the mess.