0

Good day!

I see strange behaviour of my servers. The inode table size and amount of open inodes are sharply decrease (from 90k to 10k in 5 minutes). At same time one of my application catch a large amount of swap. But external requests are in normal and is no heavy load.

Im am new in system administration. As for me it is very interesting, why does inode table decrease? Anybody has any ideas?

dev.brutus
  • 211
  • 3
  • 7
  • What do you mean by "inode table" and "open inodes"? Free inodes? Opened files? – Hauke Laging Apr 17 '13 at 10:45
  • It is [Munin](http://munin-monitoring.org/) terminology. I think the "inode table size" means total amount of inodes. I tell about such graph: http://serverfault.com/questions/330728/inode-table-usage-spiking-every-morning-at-8am – dev.brutus Apr 17 '13 at 11:26
  • Can you chack that with `df -i`? Is this a file system with dynamic inode creation? – Hauke Laging Apr 17 '13 at 11:32
  • This is values from `/proc/sys/fs/inode-nr`. Number of inodes the system has allocated. Description is [here](http://serverfault.com/questions/177379/monitoring-proc-sys-fs-inode-nr) – dev.brutus Apr 22 '13 at 08:11

1 Answers1

1

Each time a file or directory is created an inode is consumed. When the file/directory is deleted the corresponding inode is freed.

Something on your system has created 80K files/folders.

df -i 
Filesystem              Inodes  IUsed   IFree  IUse%  Mounted on
/dev/mapper/gubbins     1950240 131677 1818563    7%  /

touch fred
Filesystem              Inodes  IUsed   IFree  IUse% Mounted on
/dev/mapper/gubbins     1950240 131678  1818562    7% /

rm fred
Filesystem              Inodes  IUsed   IFree   IUse% Mounted on
/dev/mapper/gubbins     1950240 131677  1818563    7% /
user9517
  • 115,471
  • 20
  • 215
  • 297