How do I truncate or shrink large text file under UNIX / Linux operating systems?
There are various tools and methods to truncate large text files under UNIX / Linux operating systems as follows.
Options #1: Shell Output Redirction
Your can truncate text file and make the size to zero using redirection:
> {filename} ls -l largefile.txt > largefile.txt ls -l largefile.txt |
Please note that largefile.txt file is created if it doesn’t exist. And largefile.txt file is overwritten if it exists.
Option #2: truncate Command
Use the truncate command to shrink or extend the size of each FILE to the specified size:
truncate -s 0 {filename.txt} ls -lh filename.txt truncate -s 0 filename.txt ls -lh filename.txt |
The -s option is used to set SIZE to zero. See truncate command man page for more details:
man truncate
Option #3: logrotate Utility
logrotate command is designed to ease administration of systems that generate large numbers of log files. It allows automatic rotation, compression, removal, and mailing of log files. Each log file may be handled daily, weekly, monthly, or when it grows too large. See how to use logrotate command to rotates, compresses, and mails system logs stored in /var/log and other locations under UNIX / Linux oses.
Option #4: /dev/null
The null device /dev/null act as the black hole that discards all data written to it under Unix like operating system. You can use it as follows (hat tip to Philippe Petrinko):
cp /dev/null largefile.txt |
OR
cat /dev/null > largefile.txt |
23 comment
Isn’t this more of a useless thing ? You lose all the data in the first two methods.
@Nilesh,
That is the point, you want to truncate file size. For example, some stupid app crate a log file due to some networking issue. Next you fixed the issue and you don’t want the data. So just truncate it.
I vote for Option#3. Can’t live without my logrotate.conf.
Jaysunn
Option #4:
cp /dev/null toobig
Option #5:
cat /dev/null > toobig
Option #4:
cp /dev/null toobig
Option #5:
cat /dev/null > toobig
What does it mean to “rotate” log files?
@tip
http://linux.die.net/man/8/logrotate
“Please note that largefile.txt file is created if it doesn’t exist. And largefile.txt file is overwritten if it exits.”
Last word should be “exists” :)
Thanks for the heads up!
Truncating brings up a good time to remind users about sparse files. Before you try to truncate a file, make sure the file isn’t a sparse file. With sparse files, checking the size with ls -l will NOT give you the true file size. In fact, the file may look many times larger than it really is.
Always check the file with the command du to check the size of the file if there is any chance you may be dealing with a sparse file or if you aren’t sure.
@Dave: Yes, good to know/remember.
Nevertheless, [ls] can show blocks with [s / –size] option:
dd if=/dev/zero of=sparse bs=1 count=1 seek=1024k # let's create a sparse-file
ls -ls sparse # show me your blocks
8 -rw-r--r-- 1 user1 user1 1048577 feb 24 18:38 sparse
mentioning the ‘truncate’ command (as opposed to the library call), makes this Linux only (or other Unix-like without its own file utils, relying on GNU).
Please execute the below command in shell
> filename
it will truncate the file..
Thanks,,
Rama
BTW Vivek, when will you add my options, as they are functional and relevant?
And yes, please modify “exits” into “exists” as Shoaibi pointed out, don’t you think?
Done. The faq has been updated.
I appreciate your feedback :)
Rama Akella points out the useless use of cat . The end goal is that you don’t want to change the inode of the file you wish to truncate to zero bytes. It’s functionally equivalent to run `> file` to accomplish this.
http://partmaps.org/era/unix/award.html#cat
The truncate option that truncates the file in place saves me from the problem of having to dd the file to a temp file and write it back as I’m dealing with files in the hundreds of gigabytes. My outdated install of Gentoo doesn’t provide truncate in coreutils. I instead found the source from a FreeBSD machine and compiled it manually, not knowing at the time that the newer coreutils includes it.
Locate the truncate.c file in the FreeBSD world source tree. I found it here:
/usr/src/usr.bin/truncate/truncate.c
Copy this to your *nix system and compile it
gcc -o truncate truncate.c
Test it out in case the C libraries are too divergent before you use it.
For Gentoo I created an ebuild for those who are not able to update coreutils. I offer absolutely no warranties of my work or the files that I am using. I am redistributing the code without modification.
Good Article and simple topic.
Is there a yum package I’m missing?
bash: truncate: command not found
Roger, I made a post about this issue, http://www.devunmounted.com/2012/10/22/truncate-your-proble/ , I hope it helps you with your problem. The jest is that the truncate ships with a newer glibc, but if you don’t have a newer glibc you can either compile truncate.c from the newer glibc or from a libc bsd source. I believe my write up on it should cover both. Good luck.
I believe ‘truncate’ means to shorten, not (necessarily) to empty.
I would like to shorten a very large log file to, say, 1MB.
I wonder if Philippe P’s method demonstrates this? Thx!
@jangyman
Why would you do that?
What part of the original file would you keep? The first 1Mbytes, or the last 1MB.
Assuming you do not care whatever is still in your log file, (otherwise you would not consider loosing data by truncating it), your could :
# first empty it
cp /dev/null /somedir/mylogfile
# enlarge your log file to 1 MB
dd if=/dev/zero of=/somedir/mylogfile bs=1M count=1
What do you think?
You’d be better off typing
truncate -s 1000000 /var/log/*log
i usually do something like
truncate -s 24000 /var/log/*log
but my router is the linux machine, so 8 gigs is all i have of data. and some people only have 2-4 gigs.
even shorter:
:> toobigfile