Shell Programming and Scripting

Hi,
Using solaris 10 5.10 o/s
I am learning awk as I work here on the job.
What I need to do is isolate the trace file for the last 24 hours. After that I need to open those trace files and search for 'TNS-|ORA-' message from each one. These trace files MAY HAVE an occurance of them. I only need to recognize either one just once above before moving onto the next file otherwise it will become time consuming. The ultimate goal here is to do a total count (wc -l) for metric purposes. In other words if that line item includes the above it is to be counted.
So far this is what I got and would like to have some assistance or guidance as to how to complete this.

Code:

ls -l /oracle/diag/rdbms/*/*/trace/*d00*.trc | grep 'Jan 20' | awk '{print $9}'
output:
/oracle/diag/rdbms/abcdef/abcdef/trace/abcdef _d000_21750.trc
/oracle/diag/rdbms/abcdef/abcdef/trace/abcdef_d000_22001.trc
/oracle/diag/rdbms/abcdef/abcdef/trace/abcdef_d001_22002.trc

I generated the above list.
This is what I believe needs to happen next:
I need to loop thru this list, open each line using ‘more’ or ‘cat’ and do a "find" on 'TNS-|ORA-'. If it returns positive add it to the count.
count wc -l.

This part doesn’t work:

Code:

| awk 'BEGIN {more *.trc;} {egrep 'TNS-|ORA-' *.trc} END | wc –l

Any help or suggestions in this matter would be appreciated.
Thanks
al

Moderator's Comments:

Posting "Does not work" without explanation does not help you or anyone. If a command does not work for you, please show the exact circumstances you used it, and the exact error or malfunction you received. Do not paraphrase errors, or post the text as links, images, or attachments if you can avoid it: Paste the exact message, in code tags, like [code] text [/code] or by selecting the text and using the button. Thank you. The UNIX and Linux Forums

Not the most efficient solution, but your code could probably use a rewrite from scratch. The ls -l is especially pointless and likely to break down.

Code:

ls -l /oracle/diag/rdbms/*/*/trace/*d00*.trc | grep 'Jan 20' | awk '{print $9}' | 
        xargs awk '/TNS-|ORA-/' | wc -l

Try this:

Code:

find /oracle/diag/rdbms/*/*/trace -type f -name '*d00*.trc' -mtime 0 -exec egrep -c 'ORA-|TNS-' '{}' '+'

Without further info I can tell you that this:

Code:

awk 'BEGIN {more *.trc;} {egrep 'TNS-|ORA-' *.trc} END | wc –l

cannot work. You have some logical and syntactical errors in there, e.g. a missing closing quote and awk and shell mixed up. awk does not know more or egrep, it needs a command or a script built from pattern {action} pairs. action is one or more awk commands, it cannot cope with shell commands.

That can be simplified to

Code:

ls -l /oracle/diag/rdbms/*/*/trace/*d00*.trc | awk '/Jan 20/ {print $9}' | 
        xargs egrep -c 'TNS-|ORA-'

Hi ,

Here is the final solution that works for me. I spent 3 hours learning about the syntax and making adjustments. This I am presently using at work.

Code:

find /oracle/diag/rdbms/*/*/trace -type f -name '*d00*.trc' -mtime 0 -exec egrep -c 'NS Primary Error' '{}' '+'

After reviewing the results I made an adjustment to the search string to 'NS PRIMARY ERROR' . This returned unique timestamps when the files were detailed indicating the total number of times the request was refused.

Thanks for your suggestions.

regards

al