inode table overflow

[chanatip]

I got below messages. Is it possible to be cause of server hang?

vmunix: vxfs: WARNING: msgcnt 1 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 1 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 2 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 2 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 3 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 3 mesg 014: V-2-14: vx_iget - inode table overflow
vmunix: vxfs: WARNING: msgcnt 4 mesg 014: V-2-14: vx_iget - inode table overflow

[Patrick Wallek]

From what I am seeing from teh HP knowledge base, it appears that your vx_ninode kernel parameter may be set too small.  

A value of '0' for vx_ninode means that the system will determine what the value should be.  If you wish to set it differently then the minimum value is 16384 (per the vx_ninode man page on an 11.31 server).

To determine the current value:

On HP-UX 11.31 and 11.23:

# kctune vx_ninode

[chanatip]

I just increased it. But I would like to know if it's too low, it possible to be cause of server hang or not? Thank you.

# kctune vx_ninode
Tunable Value Expression Changes
vx_ninode 40000 40000 Immed

# kctune vx_ninode=131072
==> Update the automatic 'backup' configuration first? once
* The automatic 'backup' configuration has been updated.
* Future operations will ask whether to update the backup.
* The requested changes have been applied to the currently
running configuration.
Tunable Value Expression Changes
vx_ninode (before) 40000 40000 Immed
(now) 131072 131072

# kctune vx_ninode
Tunable Value Expression Changes
vx_ninode 131072 131072 Immed

[Patrick Wallek]

It may be possible that the inode cache filling up could cause the system to hang.  If a new process starts and cannot add needed entry into the inode cache, then I can see what that would cause an application to hang or cause the system to appear to hang.

[Ajay Agarwal]

> I just increased it. But I would like to know if it's too low, it possible to be cause of server hang or not? Thank you.

You can monitor the inode usage using vxfsstat command.

"/opt/VRTS/bin/vxfsstat -i / " command will provide you the inode cache statistics. If the inode usage is high, it's recommended to increase it.

And yes, if it's too inode cache is too low, it may cause system hang.