Korn Shell script need to add another search criteria

[jlaw10]

I created the following script that searches the 'RMAN-00571' text in our RMAN logs and writes the results to a logfile and then will email this log to our DBAs. I want to add a search that will verify that the "Recovery Manager complete" text also exists in all the logfiles. If this string does not exist then I also want to write the filename of the file that does not include this text to the RMAN_error.log

Note: The "Recovery Manager complete" verifies that the backup job finished. So if it did not finish, the DBAs need to be alerted.

#!/bin/ksh
#
# Report RMAN errors and send email to DBA - JAO 10/31/2005
#

export DBALIST="xyz@hotmail.com,xyz2@hotmail.com"
export date=`date`

grep -i "RMAN-00571" /u002/LOG/*/* > u002/LOG/RMAN_error.log

mailx -s "${date} RMAN errors" $DBALIST < /u002/LOG/RMAN_error.log

 

[PHV]

Add the following line before the mailx:
grep -ic "Recovery Manager complete" /u002/LOG/*/* | sed -n 's!:0$!!p' >> u002/LOG/RMAN_error.log

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ181-2886

 

[jlaw10]

Thanks that worked! Preciate it!

One more thing - what does the " sed -n 's!:0$!!p' " do?

 

[PHV]

man sed

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ181-2886

 

[BrianAtWork]

jlaw,

The sed command will only print the lines that end with :0, but first it will strip the :0 out of the line.

Maybe this will help:

$ echo test:0 | sed -n 's/:0$//p'
test
$ echo test:03 | sed -n 's/:0$//p'
$ echo test:0test | sed -n 's/:0$//p'
$ echo :0test | sed -n 's/:0$//p'
$ echo :0 | sed -n 's/:0$//p'

$

(Noticed I converted the ! to / - sed isn't picky. We could also use |,#,etc.)

The -n switch tells sed not to print out any output that it normally would (the opposite switch is -e).

The s at the beginning tells sed this will be a substitution.

The :0$ inside the first set of / / tells sed to look for a :0 at the end of a line ($ means the end, ^ means the front).

The // is the part where you can specify what to substitute - in this case it is nothing - you are effectively removing what you told sed to search for. If the line was sed -n 's/:0$/:1/p' then it would replace :0 with :1

The p at the end tells sed to only print the substituted text if a substitution was made.