soroban-rpc: Check DB on health endpoint #449
Conversation
| return HealthCheckResult{Status: "healthy"} | ||
| func NewHealthCheck(reader db.LedgerEntryReader) jrpc2.Handler { | ||
| return handler.New(func(ctx context.Context) (HealthCheckResult, error) { | ||
| if _, err := reader.GetLatestLedgerSequence(ctx); err != nil { |
There was a problem hiding this comment.
sorry for the late comment but I was thinking that it might be better to declare soroban-rpc healthy if the latest ledger close time is relatively recent (e.g. within 20 seconds) because we have seen a bug where soroban-rpc is able to process the history archive snapshot and commit that to the db but then it fails to the prepare range operation on captive core. in that scenario we would have a latest ledger sequence in the db but it could be stale
There was a problem hiding this comment.
does that threshold of 20 seconds depend on checkpoint acceleration also, suggestion sounds like a worthwhile internal enhancement if the system can determine the best threshold internally, i.e. don't need to provide externalized config parameter?
There was a problem hiding this comment.
yeah, making the threshold configurable makes sense
What
Make sure that the DB is ready in the
getHealth()endpointWhy
soroban-rpcis not really ready until we reach the first checkpoint (before which the DB won't be ready, and ingestion won't be able to start)Known limitations
The
getHealthendpoint now becomes a bit heavier