Deadlocks on fresh MariaDB with high concurrency #2463
Description
Shlink version
4.4.6
PHP version
8.4.5
How do you serve Shlink
Docker image
Database engine
MariaDB
Database version
10/11
Current behavior
When transactions deadlock and require a rollback, these exceptions occur and produce an HTTP 500:
Shlink.ERROR - PDOException: SQLSTATE[40001]: Serialization failure: 1213 Deadlock found when trying to get lock; try restarting transaction in /etc/shlink/vendor/doctrine/dbal/src/Driver/PDO/Statement.php:55
Next Doctrine\DBAL\Driver\PDO\Exception: SQLSTATE[42000]: Syntax error or access violation: 1305 SAVEPOINT DOCTRINE_2 does not exist in /etc/shlink/vendor/doctrine/dbal/src/Driver/PDO/Exception.php:24
Stack traces:
Expected behavior
No HTTP 500.
Minimum steps to reproduce
Compose.yaml:
Run:
docker compose up -d db
docker compose logs -f db
Wait for db to settle, ctrl-c
docker compose up -d app
docker compose logs -f app
See if shlink is up, ctrl-c
docker compose run --build crasher
Output will be like:
(... build ...)
New configuration template added to /tmp/.siege
Run siege -C to view the current settings in that file
{ "transactions": 3384,
"availability": 99.82,
"elapsed_time": 4.91,
"data_transferred": 1.14,
"response_time": 0.02,
"transaction_rate": 689.21,
"throughput": 0.23,
"concurrency": 15.87,
"successful_transactions": 3384,
"failed_transactions": 6,
"longest_transaction": 0.22,
"shortest_transaction": 0.01
}
It seems the bug is more pronounced when the database is a bit busy growing files. The crasher therefore truncates short_urls on every start. This is on a system with fast NVMe. I guess it will be far worse on slower storage.