dgraph-io/dgraph · error

while updating persisted query's schema

Error message

while updating persisted query's schema

What it means

fixPersistedQuery rewrites the schema entry for dgraph.graphql.p_query (changing the tokenizer/indexing of the persisted query hash field) by putting a new schema badger entry. This error wraps a badger Write failure when writing that schema entry via the update() helper.

Source

Thrown at upgrade/change_v21.03.0.go:415

	}

	// Update the tokenizer in the schema.
	su := pb.SchemaUpdate{
		Predicate: x.AttrInRootNamespace("dgraph.graphql.p_query"),
		ValueType: pb.Posting_STRING,
		Directive: pb.SchemaUpdate_INDEX,
		Tokenizer: []string{"sha256"},
	}
	data, err := proto.Marshal(&su)
	if err != nil {
		return err
	}
	entry := &badger.Entry{}
	entry.Key = x.SchemaKey(x.AttrInRootNamespace("dgraph.graphql.p_query"))
	entry.Value = data
	entry.UserMeta = posting.BitSchemaPosting
	if err := update(entry); err != nil {
		return errors.Wrap(err, "while updating persisted query's schema")
	}

	// Update the type.
	tu := pb.TypeUpdate{
		TypeName: x.AttrInRootNamespace("dgraph.graphql.persisted_query"),
		Fields:   []*pb.SchemaUpdate{&su},
	}
	data, err = proto.Marshal(&tu)
	if err != nil {
		return err
	}
	entry = &badger.Entry{}
	entry.Key = x.TypeKey(x.AttrInRootNamespace("dgraph.graphql.persisted_query"))
	entry.Value = data
	entry.UserMeta = posting.BitSchemaPosting
	if err := update(entry); err != nil {
		return errors.Wrap(err, "while updating persisted query's type")
	}

View on GitHub (pinned to 759e242be6)

Solutions

1. Check the wrapped badger error (%+v) for the root cause and address it (disk space, lock file, permissions).
2. Ensure no other dgraph/badger process has the p directory open; run the upgrade offline only when the server is stopped.
3. Re-restore the backup and rerun the upgrade if entries are corrupt.
4. Backup the p directory before rerunning so a partial upgrade can be retried cleanly.

Example fix

// before
./dgraph upgrade offline -w v21.03.2 -d ./p
// after (stop server first to release badger lock)
systemctl stop dgraph
./dgraph upgrade offline -w v21.03.2 -d ./p

// Ensure the DB is open, writable, and has space before upgrade
if err := db.Update(func(txn *badger.Txn) error { return txn.Set([]byte("__probe__"), []byte("1")) }); err != nil {
	log.Fatalf("badger not writable: %v", err)
}

func isPersistedQuerySchemaWriteErr(err error) bool {
	return err != nil && strings.Contains(err.Error(), "while updating persisted query's schema")
}

if err := upgrade.OfflineUpgradeFrom2011To2103(db); err != nil {
	log.Fatalf("%+v", err) // exposes badger write failure under this message
}

When it happens

Trigger: During OfflineUpgradeFrom2011To2103, the badger.DB.Write of the entry keyed by SchemaKey(dgraph.graphql.p_query) fails — DB closed, transaction conflict, disk full, or badger error.

Common situations: Running the offline upgrade on a restored 20.11 backup; badger database locked by another process; insufficient disk space; corrupted LSM/manifiest issues.

Related errors

• Error while updating GraphQL schema.
• while updating persisted query's type
• while upgrading persisted query
• Error while dropping depreciated preds/types.
• while upgrading cors

AI-assisted analysis of dgraph-io/dgraph@759e242be6 (2026-09-01). Data as JSON: /api/errors/807ae2af1b41bf2a. Report an issue: GitHub.