1821297 Members
3236 Online
109632 Solutions
New Discussion юеВ

dmesg command

 
SOLVED
Go to solution
augusto cossa
Frequent Advisor

dmesg command

Good Day,

I have issued the command dmesg and the result was:

/stm: file system full
/archive: file system full
/home: file system full
file: table is full

But doing bdf the /stm is at 78% and /archive at 30% and /home is at 3%

What to do? Help.

A Cossa
6 REPLIES 6
Andreas Voss
Honored Contributor
Solution

Re: dmesg command

Hi,

have a look at /var/adm/syslog/syslog.log
there you can see when the messages appears (i think your full filesystems where in the past and now freed)

Greetings

ANdrew
augusto cossa
Frequent Advisor

Re: dmesg command

Andreas,

It seams that the dmesg displys information concernig the last memory used.

Thanks a lot.

A cossa
CHRIS_ANORUO
Honored Contributor

Re: dmesg command

Hi,

Check with sar -v. If file-sz was (eg 1200/1200), then know that your file table reached its set max. at one time. You can increase this value through the kernel parameter setting nfile to a higher value.
When We Seek To Discover The Best In Others, We Somehow Bring Out The Best In Ourselves.
CKT
Advisor

Re: dmesg command

Looks like you or your users were running a program that wrote temporary data to those file systems and deleted them at the end. When a program consumes all the disk space at a given time dmesg logs it and it won't delete until you reboot your system. In your case, you should trust bdf's output.
You can't turn back the clock, but you can always rewind it again.

Re: dmesg command

Hi,
The buffer for dmesg is not very big, so you cannot see all the messages,
also there is no time-stamp, so like you have seen, there is no way
of telling when the incident was. In the manual page of dmesg(1M) it is
suggested that you emty the dmesg buffer in a logfile every ten minutes, via the crontab.
If you then change this log every day, you can use this log to see the
current dmesg messages.
Bill Hassell
Honored Contributor

Re: dmesg command

dmesg does not timestamp error messages so finding out when an error occurred will require using a cron job and the dmesg - option. An example of how to do this is found in /usr/newconfig/var/spool/cron/crontab.root for 10.20 or /var/spool/cron/crontabs/crontab.root. In this example, dmesg - is run regularly and any new messages will be appended.

BTW: The logfile name used in the example is not very intuitive (/var/adm/messages). I like to change this to /var/adm/dmesg.log for easier recognition.


Bill Hassell, sysadmin