MySQL Error 2013

Error number

2013

SQLSTATE

HY000

Symbol

CR_SERVER_LOST

Product

MySQL Server

Versions

MySQL client programs and libmysqlclient for MySQL 5.7, 8.0 and 8.4 (the mysql client adds "at line N" in batch mode)

Category

Connections & Network

Last reviewed

24 October 2026

What it means

Error 2013 means the client sent a statement but the connection broke before the full answer came back, so it does not know whether the statement finished.

The client library raises 2013 (CR_SERVER_LOST) when it did not get an error while writing to the server but did not get a full answer, or any answer, to its query. The MySQL manual gives three likely causes: network trouble while a query runs or a large result is transferred (where raising net_read_timeout can help), a failure during the initial connection (a message mentioning "reading authorization packet", where connect_timeout and the Aborted_connects status counter are the clues), and a value larger than max_allowed_packet, which some clients report as 2013 rather than 1153.

In our runs, the mysql 8.0.46 client on Windows reported 2013 whenever the server had ended its connection: when the connection was killed in the middle of a query, when it was closed for being idle before the next statement (the manual notes that on Windows the client often sees the failure only when reading the answer), and when a statement exceeded max_allowed_packet. PHP mysqlnd reported all of these as 2006. Either way, the important point is that the connection is gone and the outcome of the last statement is unknown: a write may or may not have been committed.

Common causes

• A long query or a large result set crosses a network device (load balancer, proxy, NAT gateway, VPN) that closes connections after a fixed time or when idle.
• The server ends the connection: an administrator or a tool runs KILL, the query hits a limit enforced by a proxy, or the server restarts or crashes (check the error log and Uptime).
• The client was idle longer than wait_timeout, so the next statement found the connection closed (reported as 2013 by the mysql client on Windows).
• A statement or row is larger than max_allowed_packet; the server closes the connection and some clients report 2013.
• The initial connection takes longer than connect_timeout (slow DNS, a slow or distant network), giving a message about reading the authorization or initial packet.

Diagnose it

Read-only These statements only read metadata (information_schema, performance_schema and SHOW statements) or your own data. They do not change data, privileges or configuration, so they are safe to run first.

SHOW VARIABLES LIKE 'net_read_timeout';
SHOW VARIABLES LIKE 'net_write_timeout';
SHOW VARIABLES LIKE 'connect_timeout';
SHOW VARIABLES LIKE 'wait_timeout';
SHOW VARIABLES LIKE 'max_allowed_packet';

SHOW GLOBAL STATUS LIKE 'Uptime';
SHOW GLOBAL STATUS LIKE 'Aborted_connects';
SHOW GLOBAL STATUS LIKE 'Aborted_clients';

SELECT id, user, host, db, time AS seconds, LEFT(info, 100) AS statement_start
FROM performance_schema.processlist
WHERE command = 'Query'
ORDER BY time DESC
LIMIT 10;

SELECT error_number, error_name, sum_error_raised, last_seen
FROM performance_schema.events_errors_summary_global_by_error
WHERE error_number = 1153;

How to fix it

Each fix is labelled with what it changes. Try read-only checks and the least invasive fix first, and test changes outside production.

SELECT id, customer_id, total
FROM orders
WHERE id > 50000
ORDER BY id
LIMIT 1000;

SET SESSION net_read_timeout = 120;
SET SESSION net_write_timeout = 300;  -- this connection only, in seconds

Examples

The server closes an idle connection before the next statement

Run as a script with the mysql 8.0.46 client on Windows. The client's system command (available on Windows from MySQL 8.0.19) pauses it for about 4 seconds; use system sleep 4 on Linux. PHP mysqlnd reports the same idle timeout as 2006.

SET SESSION wait_timeout = 2;
system ping -n 5 127.0.0.1 > NUL
SELECT 1;

ERROR 2013 (HY000) at line 3: Lost connection to MySQL server during query

A script that keeps going after its connection was killed

Run with mysql --force, which continues after errors. The KILL itself reports ERROR 1317 (70100) at line 1: Query execution was interrupted; every later statement then fails with 2013, because the batch-mode client did not reconnect.

KILL CONNECTION_ID();
SELECT 1;

ERROR 2013 (HY000) at line 2: Lost connection to MySQL server during query

The connection is killed while a query is running

Captured by running the mysql 8.0.46 client in batch mode with SELECT SLEEP(10) and ending that connection from a second connection of the lab. The lab cannot run a second session next to the client, so this one is not rerun automatically. KILL QUERY instead of KILL CONNECTION gave no error: SLEEP() returned 1 and the script continued.

-- Session A (mysql client):
SELECT SLEEP(10);

-- Session B, while A is still running:
KILL CONNECTION 1234;  -- A's connection id

ERROR 2013 (HY000) at line 1: Lost connection to MySQL server during query

A statement larger than max_allowed_packet

Captured with the mysql 8.0.46 client against a server with max_allowed_packet 64 MB. The server rejected the packet (error 1153) and closed the connection, and this client reported 2013, as the manual says some clients do. The statement is too large to store in the page, so the lab does not rerun it.

-- A file holding one statement of about 70 MB, run with a large client-side limit:
-- mysql --max-allowed-packet=128M < big_insert.sql
SELECT LENGTH('xxxx...about 70 MB of text...') AS len;

ERROR 2013 (HY000) at line 1: Lost connection to MySQL server during query

How to prevent it

Keep individual statements and transfers bounded: page large reads, load data in chunks, and index long-running queries. Make sure every timeout on the path (pool idle time, proxy and load balancer timeouts, wait_timeout, net_write_timeout) fits the longest legitimate query and that the shortest one is in the client, where you can handle it.

Treat a lost connection as a normal event: reconnect, check whether the last write happened, and retry idempotent work. Alert on server restarts and on rising Aborted_connects or Aborted_clients so a network or server problem is found early.

Common questions

Learn more on CodeWithSQL

• Transactions, locks and deadlocks
• Best MySQL monitoring tools

Official documentation

• Lost connection to MySQL server
• MySQL server has gone away (CR_SERVER_GONE_ERROR and CR_SERVER_LOST)
• MySQL client error message reference
• mysql client commands (system)

Last technically reviewed 24 October 2026; examples and diagnostics run on MySQL 8.0.46. Message templates checked against the server's error catalogue (perror) for that version.