Unix admin question - file system is filling up

Question

Oracle 9.2.0.1.0
UNIX AIX 5.2
The oracli3 filesystem was hitting the upper 90%s and I informaed the DBA. Learning from him is just leaning over his shoulder scribbling down as much as I can see while he thrashes through it.
I gleaned some things today that I wanted to check with tthe UNIX pros out there.
As I explained, that particular filesystem was filling up, and so he made some searches, first in the /patches directory. There were not too many there (in size), so he next went to /oracli1/oracle/iAS and did a search for logs like this:
sudo find . -name '*.log' -ls
He found one huge log file, and so had me do the cat /dev/null > command to send it to the black hole.
My question is this - when I next see that we are filling up as it were, should I automatically look for logs? Are they the culprits some of the time? Can offending logs be dealt with in this way (cat /dev/null >) ?
Is there something else I should be looking for ? (The oracli3 is now at 93% so I think we still have things to look at)
Many thanks.
DA
Sorry - one other thing - when I use the
find . -name '*.log' -ls command, how can I change it so that the output is more like that of the "more" suffix? The problem is that I would like to scroll through the output of .log files that this command throws back at me.
Message was edited by:
Dan A

Answer 1

It all depends.
When you assume there's nothing useful in those log files, you hire an Unix admin with zero knowledge of Oracle and have hem wipe those files away, 'because the file system fills up'. In the past we had similar administrators throwing away archived redo log files for exactly the same reason.
YOU on the other hand should set up proper administration procedures to move them away, tar them, and compress the tar file, so they take less space. After another 30 days you throw away the compressed tars.
What you have now is a very unprofessional method to 'administrate' a server.
Actually you are not administrating it all, because when disaster strikes, or disaster stroke in the past, and you need to explain and find out what happened and especially WHY, you will be yelling at the Unix 'admin' because he just has thrown away all log files, 'because the disk was getting full'
As to the second question:
There are a few fundamental concepts in Unix
1 ALL output can be redirected using > < and |, The latter symbol is called the pipe symbol and feeds the output into a filter. The filter is a command which reads the output from a previous command and processes. More, pg, and sort are filters, even sqlplus can be a filter
2 NO special switches should be implemented in any command if you can accomplish it using redirection or filters. So you won't find
TYPE/PAGE (VMS, Dos), but you will find cat <filename> | more
So the answer is you need to add | more to the command line
find . -name '*.log' -ls | more
You would also be advised to try to follow an Unix course, as your question is very fundamental. Alternatively buy the book Classic Shell scripting
Sybrand Bakker
Senior Oracle DBA