feat(store): add WaitForStoreVersion API to wait for cluster store replication up to target version

This commit is contained in:
Pasha Sviderski committed 2026-09-08 10:52:53 +10:00
1 parent c6d0371f15
commit 0f6c03240b
9 files changed
+450 -134

No files matched your search

+106
View File
@@ -0,0 +1,106 @@
package store
import (
"context"
"errors"
"fmt"
"strings"
"time"
"github.com/google/uuid"
)
// ErrInvalidStoreVersion indicates an invalid actor UUID in a store version vector.
var ErrInvalidStoreVersion = errors.New("invalid store version")
// WaitForVersion waits until the local store has reached each actor's minimum version in minVersion, with no known
// missing or pending transactions through those versions. Corrosion may satisfy a version by applying its surviving
// changes or by marking it complete because its changes have been superseded.
//
// Waiting normally makes the captured data available locally. However, if another write replaces some of that data
// before it arrives, Corrosion can complete the older version without transferring the replaced data. If the
// replacement is outside minVersion, this method can return while the affected data is still missing or outdated.
// This can happen during concurrent updates even when all machines are well connected.
//
// This method does not guarantee an exact snapshot or delivery of every historical value.
// Callers that require a specific record or condition should verify it after waiting.
//
// The method observes native replication without initiating synchronisation.
// An empty minVersion requires no replication. The context controls cancellation and the deadline.
func (s *Store) WaitForVersion(ctx context.Context, minVersion map[string]uint64) error {
if err := ctx.Err(); err != nil {
return err
}
placeholders := make([]string, 0, len(minVersion))
args := make([]any, 0, 2*len(minVersion))
for actor, version := range minVersion {
actorID, err := uuid.Parse(actor)
if err != nil {
return fmt.Errorf("%w: actor '%s' is not a UUID: %v", ErrInvalidStoreVersion, actor, err)
}
if version == 0 {
continue
}
// Corrosion accepts a JSON array of bytes as a BLOB parameter. A UUID marshals as a string,
// and a byte slice marshals as base64, so use a plain fixed-size array instead.
placeholders = append(placeholders, "(?, ?)")
args = append(args, [16]byte(actorID), version)
}
if len(args) == 0 {
return nil
}
// Check all bookkeeping in one SQLite snapshot. Raw buffered rows can remain after application,
// but sequence bookkeeping is removed in the same transaction that applies or clears a version.
query := `WITH min_version(actor_id, version) AS (VALUES ` + strings.Join(placeholders, ", ") + `)
SELECT NOT EXISTS (
SELECT 1 FROM min_version
LEFT JOIN crsql_db_versions AS current ON current.site_id = min_version.actor_id
WHERE COALESCE(current.db_version, 0) < min_version.version
OR EXISTS (
SELECT 1 FROM __corro_bookkeeping_gaps
WHERE actor_id = min_version.actor_id AND start <= min_version.version
)
OR EXISTS (
SELECT 1 FROM __corro_seq_bookkeeping
WHERE site_id = min_version.actor_id AND db_version <= min_version.version
)
)`
ticker := time.NewTicker(100 * time.Millisecond)
defer ticker.Stop()
for {
rows, err := s.corro.QueryContext(ctx, query, args...)
if err != nil {
return fmt.Errorf("check store replication: %w", err)
}
if !rows.Next() {
if err = rows.Err(); err != nil {
return fmt.Errorf("check store replication: %w", err)
}
return errors.New("check store replication: store version query returned no rows")
}
var reached int
if err = rows.Scan(&reached); err != nil {
rows.Close()
return fmt.Errorf("check store replication: %w", err)
}
rows.Next() // Consume the end-of-query event.
rows.Close()
if err = rows.Err(); err != nil {
return fmt.Errorf("check store replication: %w", err)
}
if reached == 1 {
return ctx.Err()
}
select {
case <-ctx.Done():
return ctx.Err()
case <-ticker.C:
}
}
}