Resource Usage
Disk Usage
Note: If you have a file/files open and your disk space appears full, do a reboot and it will close the open file/s. This is a true indication of disk space then.
To check overall disk usage run:
du -h df -h
Use the below command to show number of files in target directory:
ls -1 target|wc -l
For Inodes:
df -i find . -xdev -type f | cut -d "/" -f 2 | sort | uniq -c | sort -n cd /directory that is the largest #run find command again
Disk check
apt-get update apt-get install smartmontools smartctl -H /dev/sda smartctl -A /dev/sda
This will check the health of the hard drive. If you use RAID the software may not be able to pick it up. If you are getting errors, check you know what your HDD type you have by doing:
smartctl --scan
Do not trust df, to check the HDDs use:
fdisk -l
If your server uses SCSI, you may need to the following each time:
smartctl -d scsi /dev/sda
A better alternative is mdadm but be careful, do not install it unless you know what you are doing as it will prompt you with questions via a GUI straight away.
Empty a file
:> /file/path
This always exits true, but produces no output. Good for large old log files.
More complicated options
- Changing the second n option specifies the amount of results you want to show.
du --max-depth=1|sort -nr|head -n10 #this prints out the largest value at the top du --max-depth=1|sort -n|tail -n5 #this prints out the largest value at the bottom du --max-depth=20|sort -nr|head -n20>diskusage.txt #this outputs the top 20 results into a file du --max-depth=1|sort -nr|tail -n10 #this prints out the smallest value at the bottom, good for temp directories du --max-depth=1|sort -n|head -n10 #this prints out the smallest value at the top du --max-depth=1 -h * |sort -n -k 1 |egrep 'M|G' du -sh *
http://www.earthinfo.org/linux-disk-usage-sorted-by-size-and-human-readable/
If the server is Windows, you can use Performance monitor which is built into Windows to run on the server to monitor free space along with disk read/writes. You may wish to install http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx as it will provide a greater overview as to what is writing to the server and when.
If there is a program writing to the temporary directory it may be putting the files in there and then removing them once they are finished being used hence the fluctuation.
ncdu
A useful tool for Unix is called ncdu.
This allows you to discover the overall usage, please note this can take a while depending on the amount of data you have.
From root (cd /) run one of the following commands to install and run ncdu:
CentOS (you may need to enable the EPEL repository):
yum search ncdu yum install ncdu.x86_64 cd / #or the directory you want to check ncdu
Ubuntu:
apt-get install ncdu cd / #or the directory you want to check ncdu
du -h --max-depth=1 | grep G du -kh | grep G du -khs * | grep G
- Using grep looks down one directory and looks for GB disk usage, use M for MB, K for KB.
- To dive down more directories change the max depth value.
Exlcuding a directory in ncdu:
ncdu --exclude /home/mc/Minecraft/plugins/dynmap/web/tiles/world
General Usage
On Linux:
top apt-get install htop #or yum install htop #then htop
apt-get install goaccess
Also look at http://serverkb.co.uk/wiki/Scripts
On Windows use Task Manager and Performance Monitor.
Munin and Graphite are good for system management.
Memory available
Display the top ten running processes - sorted by memory usage:
ps aux --sort -rss | head
free -m
This shows you the available and currently used memory (RAM) in MB. So 1900-2000 is 1-2GB. On a Shared/VPS environment normally half the Total and you get your Soft Limit (guaranteed), the full amount is your Hard Limit (burstable, not guaranteed).
cat /proc/user_beancounters
You will see something like this:
Version: 2.5
uid resource | held | maxheld | barrier limit | failcnt
328178: kmemsize | 4348996 | 4366996 | 16384000 18022400 | 0
lockedpages | 0 | 0 | 4096 4096 | 0
privvmpages | 68233 | 68262 | 153600 153600 | 34995
or for example on an 8GB of RAM server:
uid resource | held | maxheld barrier limit | failcnt
506378: kmemsize | 62055413 | 62490301 9223372036854775807 9223372036854775807 | 0
lockedpages | 0 | 0 9223372036854775807 9223372036854775807 | 0
privvmpages | 384054 | 384315 9223372036854775807 9223372036854775807 | 0
shmpages | 1687 | 1687 9223372036854775807 9223372036854775807 | 0
shmpages | 1687 | 1687 9223372036854775807 9223372036854775807 | 0
dummy | 0 | 0 9223372036854775807 9223372036854775807 | 0
numproc | 205 | 208 9223372036854775807 9223372036854775807 | 0
physpages | 119511 | 119641 9223372036854775807 9223372036854775807 | 0
vmguarpages | 0 | 0 9223372036854775807 9223372036854775807 | 0
oomguarpages | 119513 | 119643 9223372036854775807 9223372036854775807 | 0
numtcpsock | 34 | 36 9223372036854775807 9223372036854775807 | 0
numflock | 21 | 23 9223372036854775807 9223372036854775807 | 0
numpty | 1 | 1 9223372036854775807 9223372036854775807 | 0
numsiginfo | 4 | 5 9223372036854775807 9223372036854775807 | 0
tcpsndbuf | 601840 | 648400 9223372036854775807 9223372036854775807 | 0
tcprcvbuf | 557056 | 589824 9223372036854775807 9223372036854775807 | 0
othersockbuf | 2132376 | 2133680 9223372036854775807 9223372036854775807 | 0
dgramrcvbuf | 0 | 2608 9223372036854775807 9223372036854775807 | 0
numothersock | 947 | 949 9223372036854775807 9223372036854775807 | 0
dcachesize | 3301507 | 3312775 9223372036854775807 9223372036854775807 | 0
numfile | 20555 | 20630 9223372036854775807 9223372036854775807 | 0
numiptent | 15 | 15 9223372036854775807 9223372036854775807 | 0
Take a look at the numbers left to the first Pipe (|) in the held column: These stand for the actual usage. Right of the last pipe: Failcounts. Having failcounts inidicates that a service is consuming all the ram.
