UNIX for Dummies Questions & Answers

Hi,

I'm currently using the following command to wipe clean a log file which can't be straight out RM'd:

Code:

cat /dev/null > server.log

I'm building this into a script and I'm current working on a command to run on each machine to do this automatically however I have multiple files so I need to do these all at the same time by running a check using 'find' and listing files above 10G.

Code:

find /data/logs/ -name server.log -size +10G | xargs cat /dev/null >{}

Looking for a solution similar to the above that will pass /dev/null to the file, blanking it essentially.

Any help is appreciated.

How sure are you that this will work? Truncating an open file is not guaranteed to release the space it uses, and the process(es) that already have the file open could very well just write more data at their current offset, which at best will create a sparse file.

The only way to make something like this work is if you have control of the application(s) generating the log file(s).

And if you have control of the application(s), you don't need a kludge like this because you'd have the app(s) properly manage their log files.

It seems to have worked previously from what I've seen when running the command manually, so I see no reason why it wouldn't work now.

Any suggestions?

Code:

cat /dev/null > something

magic power is kind of a urban legend. "cat /dev/null" does pretty much nothing useful and can be replaced by the null or even an empty command, i.e.

Code:

: > server.log

or simply

Code:

> server.log

Beyond that, I concur with achenle it is an unreliable way to zero out an open log file.

how about doing it all in find instead of piping to xargs...

Code:

find /data/logs/ -name server.log -size +10G -exec cp /dev/null {} \;

thogh i am not sure if 10G is a valid argument to -size.

Again, there is no magic with /dev/null, whether it is used with cat or cp.

Here is something both optimized and a little cryptic :

Code:

find /data/logs/ -name server.log -size +10G -exec sh -c 'for i; do > $i; done' {} +

"+10G" is a GNU find extension, otherwise, "+2097152" would equally work and with any standard compliant find.

And so is this...

Code:

find /data/logs/ -name server.log -size +10G -exec cp /dev/null {} +

addition of the trailing + prevents the shell from spawning a new process for every file