When configuring logrotate to trigger based on file size (size or maxsize), many administrators encounter the same fundamental misunderstanding: logrotate itself doesn't monitor files continuously. The rotation only occurs when logrotate is executed, typically through cron.

Your current configuration has two competing directives:

daily       # Run rotation daily
size 250M   # Rotate if over 250MB

The daily parameter means the rotation will only be attempted once per day during the normal logrotate cron job (usually in /etc/cron.daily/logrotate). At that time, it will check both conditions.

For true size-based rotation, you need to:

1. Remove the daily parameter
2. Create a dedicated cron entry for this logfile

Example crontab entry (run every 5 minutes):

*/5 * * * * /usr/sbin/logrotate /etc/logrotate.d/haproxy -f

Here's an optimized configuration that handles both size and time-based rotation:

"/var/log/haproxy.log" {
    size 250M
    rotate 5
    create 644 root root
    missingok
    compress
    delaycompress
    copytruncate
    sharedscripts
    postrotate
        /usr/bin/systemctl reload haproxy >/dev/null 2>&1 || true
    endscript
}

This removes the daily restriction while maintaining all other functionality.

• The copytruncate method may lose some logs during rotation
• Frequent rotations increase I/O load
• Consider using dateext for clearer rotation naming

Force a test run with debug output:

logrotate -d /etc/logrotate.d/haproxy

Or perform a dry-run:

logrotate -vf /etc/logrotate.d/haproxy

---

When using logrotate with both daily and size/maxsize directives, many administrators encounter unexpected behavior. The configuration shown:

"/var/log/haproxy.log" {
  daily
  size 250M
  rotate 1
  create 644 root root
  missingok
  compress
  notifempty
  copytruncate
}

has a fundamental limitation - the daily directive takes precedence over size-based rotation. This means logrotate will only check the file once per day (typically via cron.daily) regardless of size thresholds.

Option 1: Remove Daily and Use Custom Cron

First modify your configuration:

"/var/log/haproxy.log" {
  size 250M
  rotate 1
  create 644 root root
  missingok
  compress
  notifempty
  copytruncate
}

Then create a custom cron job (e.g., in /etc/cron.d/logrotate_haproxy):

*/30 * * * * root /usr/sbin/logrotate /etc/logrotate.d/haproxy

Option 2: Use inotifywait for Real-Time Monitoring

For critical services needing immediate rotation:

#!/bin/bash
inotifywait -m /var/log -e modify |
while read path action file; do
  if [[ "$file" = "haproxy.log" ]]; then
    size=$(stat -c%s "/var/log/haproxy.log")
    if [ $size -gt $((250*1024*1024)) ]; then
      /usr/sbin/logrotate -f /etc/logrotate.d/haproxy
    fi
  fi
done

1. maxsize vs size: Both work similarly but with different triggers:
- size: Rotates when log exceeds specified size at check time
- maxsize: Rotates when log might exceed size between checks

2. The copytruncate method may lose some log entries during rotation. For zero-loss solutions, consider using sharedscripts with service restarts.

3. For HAProxy specifically, you might want to add:

postrotate
    /usr/bin/systemctl reload haproxy >/dev/null 2>&1 || true
endscript