Diagnose a slow linux box: cheat sheet

Posted by & filed under Informatique, Linux.

Disk Space:

df -h

Make sure you have enough disk space

Memory /Processes / Load average

htop
top (shift + m)
free -m

Use Htop
Check for processes that are taking up a lot of memory/CPU and Check you’re not out of memory

Apache errors

grep 500 /var/log/apache2/error.log | less

Look for 500 errors caused by erroneous code on the server

High hit rate

grep MaxClients /var/log/apache2/error.log

Check for MaxClients warningdamn in your apache error logs

Check recent logs

ls -lrt /var/log/

the -lrt flag will show you the most recently modified files at the end

Check your cronjobs

ls -la /var/spool/cron/*
ls -la /etc/cron*

crontab -l

You might find your server is going down at a certain time, this could be result of a cronjob eating up too many resources

Check Kernel Messages

dmesg

Check inodes

df -i

Check inodes remaining when you have a disk that looks full but is reporting free space

Determine how many apache threads are running (if you’re not using mod_status)

ps -e | grep apache2 | wc -l

For DOS attacks:

Start Number of active, and recently torn down TCP sessions

netstat -ant | egrep -i '(ESTABLISHED|WAIT|CLOSING)' | wc -l

Number of sessions waiting for ACK (SYN Flood)

netstat -ant | egrep -i '(SYN)' | wc -l

List listening TCP sockets

netstat -ant | egrep -i '(LISTEN)'

List arguments passed to program

cat /proc/$PID/cmdline

For DOS attacks: END

Leave a Reply