alibaba/canal · error · PositionNotFoundException

can't find start position for {}

Error message

can't find start position for {}

What it means

Thrown as PositionNotFoundException when AbstractEventParser cannot determine a binlog start position after calling findStartPosition(erosaConnection). Without a start position the parser cannot begin a binlog dump, so startup aborts. It is the most common 'fresh start with no position' failure.

Source

Thrown at parse/src/main/java/com/alibaba/otter/canal/parse/inbound/AbstractEventParser.java:189

                        preDump(erosaConnection);

                        erosaConnection.connect();// 链接

                        long queryServerId = erosaConnection.queryServerId();
                        if (queryServerId != 0) {
                            serverId = queryServerId;
                        }

                        if (erosaConnection instanceof MysqlConnection) {
                            isMariaDB = ((MysqlConnection) erosaConnection).isMariaDB();
                        }
                        // 4. 获取最后的位置信息
                        long start = System.currentTimeMillis();
                        logger.warn("---> begin to find start position, it will be long time for reset or first position");
                        EntryPosition position = findStartPosition(erosaConnection);
                        final EntryPosition startPosition = position;
                        if (startPosition == null) {
                            throw new PositionNotFoundException("can't find start position for " + destination);
                        }

                        if (!processTableMeta(startPosition)) {
                            throw new CanalParseException("can't find init table meta for " + destination
                                                          + " with position : " + startPosition);
                        }
                        long end = System.currentTimeMillis();
                        logger.warn("---> find start position successfully, {}", startPosition.toString() + " cost : "
                                                                                 + (end - start)
                                                                                 + "ms , the next step is binlog dump");
                        // 重新链接，因为在找position过程中可能有状态，需要断开后重建
                        erosaConnection.reconnect();

                        final SinkFunction sinkHandler = new SinkFunction<EVENT>() {

                            private LogPosition lastPosition;

                            public boolean sink(EVENT event) {

View on GitHub (pinned to 87be50e876)

Solutions

1. Confirm the MySQL user has REPLICATION SLAVE and REPLICATION CLIENT privileges (and SELECT on needed tables).
2. If first start, set canal.instance.master.address correctly and ensure binlogs exist; let Canal record the current master position.
3. If resuming, verify the persisted zk/file position still references an existing binlog file (not purged).
4. For GTID mode, ensure the requested GTID is present on the master and gtid mode is enabled on the source.

Example fix

-- grant the canal MySQL user the required rights
GRANT SELECT, REPLICATION SLAVE, REPLICATION CLIENT ON *.* TO 'canal'@'%';

-- ensure the canal user can report a start position
SHOW GRANTS FOR 'canal'@'%';
-- must include REPLICATION SLAVE, REPLICATION CLIENT, SELECT
SHOW MASTER STATUS; -- non-empty => a start position exists

try { parser.start(); }
catch (PositionNotFoundException e) {
    // first start or binlogs purged; seed from current master position then retry
    seedPositionFromMaster(); parser.start();
}

When it happens

Trigger: In the parse thread, after connecting and (for MySQL) reading serverId/isMariaDB, findStartPosition returns null. This occurs on first start with no persisted position, after a position reset, when the binlog does not contain a matching GTID/position, or when the configured start criteria (timestamp/journal/gtid) yield nothing.

Common situations: First launch of an instance with no cursor saved; binlogs on the MySQL master have been purged/rotated past the requested start; GTID mode but the requested GTID set is not present; timestamp-based start older than retained binlogs; MySQL user lacks REPLICATION SLAVE/REPLICATION CLIENT privileges so SHOW MASTER STATUS returns nothing.

Related errors

• binlog: + binlogfilename + is not found
• position not found!
• dir[{}] can not read/write
• use gtid and TableMeta TSDB should be config timestamp > 0
• unexpected rds serverId + serverId + in binlog file !

AI-assisted analysis of alibaba/canal@87be50e876 (2026-08-14). Data as JSON: /api/errors/7c33c264b37f3e71. Report an issue: GitHub.