/usr/adm/messages craziness

[Jonathan Biggar]

In article <659 at aplvax.UUCP> eric at aplvax.UUCP writes:
>
>	I have a VAX 11/780 running 4.1 that has decided to start dumping
>garbage into /usr/adm/messages. First off, none of this garbage is coming
>out to the console. But every ten minutes, the system appends about 40 lines
>to /usr/adm/messages, and the data is usually the same (currently a
>concatenation of a couple of boot sequences). Originally it was mostly
>a bunch of debug messages from a driver, but the debug messages have since
>been disabled and, as I said, the message is now just a couple of boot 
>sequences. Further, since this started, we have re-booted a couple of
>times. I will gladly take any recommendations on how to stop this, or what
>caused it in the first place.

Funny thing.  I had just noticed the exact same problem.  The file
/usr/adm/messages is written to by a the program /etc/dmesg which
is run every ten minutes from crontab.  The crontab entry should
look like this:

0,10,20,30,40,50 * * * * /etc/dmesg - >>/usr/adm/messages

The '-' as an argument instructs dmesg to compute the kernel error messages
incrementally, i. e. only write to /usr/adm/messages those error
messages that are new.

I thought that this would fix the problem, but our crontab was correct.

However, there is an answer!  After staring at the code for dmesg for
a while, I discovered that the file /usr/adm/msgbuf must exist for
dmesg to determine the new messages.  So there are to things you should
check:

1)  That the crontab entry is correct.

2)  That the file /usr/adm/msgbuf exists.

Jon Biggar
{allegra,burdvax,cbosgd,hplabs,ihnp4,sdccsu3}!sdcrdcf!jonab