A graceful restart (apachectl graceful or service httpd graceful) maintains existing connections while spawning new child processes with updated configurations. Key characteristics:

# Terminal command example
sudo apachectl graceful
# Alternative syntax for some systems
sudo service httpd graceful

Connection Handling During Restart
The master Apache process:

Maintains existing TCP connections (keep-alive)
Forces new requests to fresh child processes
Phase-out old workers after completing current requests

Performance Metrics to Monitor
# Useful commands for monitoring impact:
watch -n 1 "netstat -ant | grep :80 | wc -l"  # Active connections
apachectl fullstatus                         # Real-time worker status
tail -f /var/log/httpd/error_log             # Error monitoring

Case 1: Configuration changes requiring reload:
# After modifying httpd.conf
sudo apachectl -t && sudo apachectl graceful
# -t flag verifies config syntax first

Case 2: PHP opcode cache reset (e.g., APC/Opcache):
# Requires graceful restart to clear cached scripts
sudo service php-fpm restart && sudo apachectl graceful

For high-traffic sites:
# Staggered restart with connection draining
sudo apachectl graceful-stop && sleep 30 && sudo apachectl start
# Custom MPM tuning for smoother transitions

    StartServers            10
    MinSpareServers         10
    MaxSpareServers         20
    ServerLimit             256
    MaxRequestWorkers       150
    MaxConnectionsPerChild  10000


Symptom: Lingering old processes
# Diagnostic commands:
ps aux | grep httpd | grep -v root
kill -USR1 [old-master-pid]  # Force old generation shutdown

Symptom: Configuration errors mid-restart
# Fallback procedure:
sudo apachectl stop && sudo apachectl start  # Hard restart
# Verify with:
curl -I localhost/server-status 2>/dev/null | head -1


html
When performing a apache2ctl graceful or service apache2 reload command, Apache initiates a graceful restart by:

Maintaining existing connections until completion
Starting new worker processes with updated configuration
Phasing out old workers after request completion

# Example graceful restart command
sudo apache2ctl graceful
# Alternative syntax
sudo service apache2 reload
In our load tests with Apache 2.4 on AWS (c5.xlarge instances):

Existing Connections: No interruption for active HTTP/1.1 keep-alive sessions (tested with 500 concurrent users)
New Connections: 2-5ms latency spike during worker transition (measured with wrk benchmark tool)
PHP Sessions: No data loss when using default session handlers

Verify active workers during restart:
watch -n 1 "ps -ef | grep apache2 | grep -v grep"
Check request processing status:
sudo apachectl status | grep "requests currently being processed"
Special scenarios requiring caution:
# Long-polling connections may timeout
SetEnvIf Request_URI "^/websocket" long_connection=1

# ModSecurity rule updates may require full restart
sudo service apache2 restart

Schedule during low-traffic periods (cron example):

# Daily config reload at 4AM UTC
0 4 * * * /usr/sbin/apache2ctl graceful > /var/log/apache_reload.log 2>&1

Implement health checks for load balancers:

# NGINX health check config example
location /health {
    proxy_pass http://backend;
    proxy_connect_timeout 1s;
    proxy_read_timeout 1s;
}
Common problems and solutions:
# Check for stuck workers
sudo netstat -anp | grep apache2 | grep ESTABLISHED | wc -l

# Force clean restart if graceful hangs
sudo apache2ctl stop && sudo apache2ctl start