UNIX for Dummies Questions & Answers

Ravinder, I do not want to send the o/p in a seprate file. since I should test in sysctl.conf file

Hello Stew,

If I understood you correctly you are trying to take OUTPUT or EDIT the same file which is being used for input as well, then I can say there no option present in awk to do so. You can create a script and execute like following.

Code:

cat script.ksh
awk '/kernel.sem/ { if ($3 < "300") { sub($3,"300",$0); print >> "/tmp/test1" } else {print >> "/tmp/test1" }}' ACTUAL_FILE
mv /tmp/test1 ACTUAL_FILE

Hope this helps.

Thanks,
R. Singh

you are correct. I want to edit the same file which being used for i/p file and update the o/p in the same i/p file.

My requirement is, need to edit kernel.sem values from sysctl.conf file. So I can't move the file name since the sysctl file is very sensitive.

Do you know any other option.

Hello Stew,

Could you please try the same and let me know if this helps, kindly try the same first with same details a test file
and if All is Well then you can try with actual file too.

Code:

cat script.ksh
VALUE=`awk '/kernel\.sem/ {if($3 < 300){print 1}}' Input_file`
if [[ $VALUE -eq 1 ]]
then
 sed -i '/kernel.sem/s/\(.*= \)\(...\)\(.*\)/\1 300 \3/g' Input_file
else
 echo "Values are fine nothing to do."
fi

Hope this helps.

Thanks,
R. Singh

I will try this . Thanks a lot for your help

Hi,
Example standard procedure (here bash and awk):
Script:

Code:

$ cat script.bash
#!/usr/bin/bash

exec 3<demosysctl.conf
rm demosysctl.conf
awk '/kernel.sem/ { if ($3 < "300") { sub($3,"300",$0); print } ;next }1' <&3 >demosysctl.conf

Input:

Code:

$ cat demosysctl.conf
kernel.sem = 250   32000   32      128250
kernel.small = 1612364
kernel.shmmax = 2147483648

Execution:

Code:

$ ./script.bash
$ cat demosysctl.conf
kernel.sem = 300   32000   32      128250
kernel.small = 1612364
kernel.shmmax = 2147483648

Regards.

This is extremely dangerous code. Depending on the size of the input file and the location of the line being modified in the file, this may truncate your input file to size zero instead of updating a field in one or more lines in the file.

NEVER, try to read and write a file that might be larger than the size of the 1st read performed by awk or you will destroy your input file. (And, since there is no requirement placed on awk for the sizes of the reads it performs, NEVER, NEVER, NEVER do this.)

Using sed -i to update a system configuration file could also leave you with an unbootable system if your system loses power in the middle of an update. Ranvinder's suggestion in post #9 is much safer (using a temp file and moving or copying the updated file after it has been fixed), but it should verify that the awk command returned a zero exit status before attempting the mv on the next line.

However, it would be much better to create the temp file in the same directory as the file being updated (same filesystem is sufficient, but in case more file systems are created in the future, same directory is safer), so the mv is just a rename() which will always leave you with either the original configuration file or the updated configuartion file in place. (Of course, any change to a file could be corrupted by a disk block or controller going bad while you're making the update; but this method eliminates several other error cases that could destroy your data.)