Home.... | log in.... | Pub my question! | Sign-up....

Java-index >> General >> Talk to the Sysop

unexpected free inode

How bad does it sound ? Need for immediate reboot ?Feb 27 06:42:46 hostname ufs: [ID 879645 kern.notice] NOTICE: /var: unexpected free inode 400368, run fsck(1M)Feb 27 06:45:08 hostname last message repeated 5 timesThanks

[256 byte] By [NDumeige] at [2007-11-25 23:20:06]

«« automatically processing email
»» Anonymous FTP

# 1

Go to single user mode (init S), unmount /var and run fsck -y /var. This should sort your problems out. Probably a good idea to get it sorted now as opposed to later.

bmacdo at 2007-7-5 18:08:06 >

top of Java-index,General,Talk to the Sysop...

# 2

You didn't say what version of Solaris you are running but more recent versions support the "logging" mount switch.You should see if this appropriate for your environment and use it if so - assuming its not already in use.

swoneill at 2007-7-5 18:08:06 >

# 3

This just happened to me on an sds mirrored drive, Oracle redo filesystem, e3500 with sol. 9 9/02.

I had shutdown all oracle databases and rebooted the system. There were no messages indicating a problem anywhere. Upon starting, a couple of the databases complained. Further investigation showed "I/O error" when ls -l the files and then the "unexpected free inode" message in the messages file.

The system had been happily running for 3 months with no problem.

fsck'ing the /redo eventually fixed this but I am concerned over this sudden problem with no warning.

bwestb at 2007-7-5 18:08:06 >

# 4

Run iostat -e from time to time and see if errors are accumulating over time. Also, check your /var/adm/messages file and the console (assuming you have one) for messages related to disk problems.

Are the files systems Oracle is using RAW or cooked ? If cooked, you might consider using "logging" as a mount option in the /etc/vfstab file for these file systems. Actually, use logging for all your files systems.

swoneill at 2007-7-5 18:08:06 >

# 5

Hi & thank for your help,

Typing iostat -e, I get :

- errors

devices/w h/w trn tot

sd6 0101

ssd03000 30

ssd13000 30

ssd2 0000

ssd3 0000

ssd4 0000

ssd5 0000

ssd6 0000

ssd7 0000

ssd81500 15

ssd91500 15

st111410 15

nfs1 0000

What is the meaning of each field ? Manpage lacks definition for thoses.

Thanks again,

NDumeige at 2007-7-5 18:08:06 >

# 6

I prefer to use iostat -En myself, gives disk specific information.

bmacdo at 2007-7-5 18:08:06 >

# 7

When was the last time you updated your patches on this system ?Check your cable attachments. Has anyone been behind the machine in such a way that cables could have been jiggled at all ?Are you using "logging" as a mount flag for your file systems ?

swoneill at 2007-7-5 18:08:06 >

# 8

Hi,The system hase been patched recently: SunOS hostname 5.8 Generic_108528-18 sun4u sparc SUNW,Sun-Fire-880Indeed we use the logging option.You think it's some kind of cable connections problem ?Thanks

NDumeige at 2007-7-5 18:08:06 >

# 9

It could be. The software errors you are getting can easily be caused by a loss cable. I was messing around behind my errors on day and started noticing lots of software errors shortly after. Started poking around and I found a cable I had bumped. Pushed it back in and the errors disappeared.

First rule of diagnosis:

- Eliminate the hardware FIRST

swoneill at 2007-7-5 18:08:06 >

# 10

Man I can't type tonight ... LOL

swoneill at 2007-7-5 18:08:06 >

Java programming resources: FAQs, tutorials, compiler and browser download sites, documentation, books lists, IDEs, etc.

unexpected free inode

General New Topic

General Hot Topic

General

All Classified