Diagnosing 502/504 Gateway Errors

Understanding 502 and 504 Errors

A 502 Bad Gateway means OLS received an invalid response from the PHP backend (lsphp). A 504 Gateway Timeout means the PHP process took too long to respond. Both point to PHP-side issues.

Common Cause: PHP Timeout

WordPress operations like imports, plugin updates, or WooCommerce order exports can exceed the default timeout. Increase these values:

# PHP settings
max_execution_time = 300
max_input_time = 300

# OLS external app timeout (vhconf.conf)
extprocessor lsphp {
  maxConns          20
  env               PHP_LSAPI_CHILDREN=20
  initTimeout       300
  retryTimeout      0
  respBuffer        0
  backlog           100
  connTimeout       300
}

Check lsphp Processes

If all PHP workers are busy, new requests get 502 errors. Check the current worker count:

ps aux | grep lsphp | grep -v grep | wc -l

If the count equals your PHP_LSAPI_CHILDREN limit, either increase the limit or investigate which requests are consuming workers (check the access log for slow requests).

OLS LSAPI Connection Issues

If lsphp crashes or fails to start, OLS cannot forward requests. Check the OLS error log:

tail -50 /usr/local/lsws/logs/error.log

Look for messages like connection refused, lsphp is not available, or max children reached.

Restart PHP Workers

Sometimes stale lsphp processes need to be killed and restarted:

sudo killall -9 lsphp
sudo systemctl restart lsws

OLS will spawn fresh PHP workers on the next request.

Memory Exhaustion

If the server is running out of RAM, lsphp processes get killed by the OOM killer. Check:

dmesg | grep -i "out of memory"
free -h

Reduce PHP_LSAPI_CHILDREN or add swap space if memory is tight.

Quick Checklist

Check /usr/local/lsws/logs/error.log for LSAPI errors
Check /home/DOMAIN/logs/error.log for PHP fatal errors
Verify max_execution_time is sufficient
Ensure enough free RAM for PHP workers
Restart OLS if processes are stuck: sudo systemctl restart lsws