[SOLVED] how do I get a list of orphaned files ?

[magowiz]

Hi,
since my package db was corrupted I reinstalled the whole gentoo system over my root partition without deleting files.
Since now I have some files that belongs to old installation ( packages merged before the db corruption and not merged in the new installation) I would like to get a list of theese files to :
reinstall packages I forgot to reinstall but that I need or simply remove files of packages that I don't need anymore .
Is there a way to get such a list ?

[V-Li]

Code: Select all

 find / -type f | xargs qfile -o

May take a while and maybe you will exceed your command line length limit, adjust the argument of find and test the hierarchies separately.

[magowiz]

Code: Select all

 find / -type f | xargs qfile -o

May take a while and maybe you will exceed your command line length limit, adjust the argument of find and test the hierarchies separately.

thank you very much , now I'm running it in /usr and I found some files

[jongeek]

V-Li is spot on, but I check my box periodically for orphans, and wrote a little script to help me out. You can check it out here: http://pastebin.com/m22712455.

The excludes and search variables control what is looked at and what is not. Very handy for running from cron once a month or so.

[V-Li]

V-Li is spot on, but I check my box periodically for orphans, and wrote a little script to help me out. You can check it out here: http://pastebin.com/m22712455.

The excludes and search variables control what is looked at and what is not. Very handy for running from cron once a month or so.

Please don't pastebin such scripts, as they may get lost over time. For the archive here is a paste:

Code: Select all

#! /bin/bash                       

echo "start run @ " $(date "+%Y-%m-%d %H:%M") | tee /tmp/orphan-starttime

file01=/tmp/orphan01
file02=/tmp/orphan02

rm -f "${file01}" "${file02}"

search="/"
#search="/usr/share/doc/"

exclude="/bak /boot"
exclude="${exclude} /chroot /etc/gconf /etc/runlevels"
exclude="${exclude} /etc/ssl/certs"                   
exclude="${exclude} /lib/modules"                     
exclude="${exclude} /lost+found"                      
exclude="${exclude} /opt/wxwidgets /portagecrap /root"
exclude="${exclude} /usr/lib/perl5/site_perl /usr/lib/python2.5/site_packages"
exclude="${exclude} /usr/local/src"                                           
exclude="${exclude} /usr/portage"                                             
exclude="${exclude} /usr/share/mime"                                          
exclude="${exclude} /usr/src"                                                 
exclude="${exclude} /var"                                                     
exclude="${exclude} *.pyc *.pyo *.keep"                                       

predicate=""

for dir in ${exclude}
do                   
        predicate="${predicate} ${predicate:+"-o"} -path ${dir} -prune"
done

echo "building filelist ..."
echo find ${search} -xdev ${predicate} -o -print
     find ${search} -xdev ${predicate} -o -print > ${file01}

count=$(wc -l < ${file01})

echo "scanning ${count} files for orphans"

# This version gives progress information but is slow
#
#let i=1
#for file in $(cat ${file01})
#do
        #echo -ne "Checking ${i} of ${count} files...\r"
        #let i=$((i+1))
        #qfile -o "${file}" >> ${file02}
#done

# This version gives no progress information but is faster
#
echo "qfile'ing ..."
qfile -o -f ${file01} > ${file02}

echo
echo
echo "Found $(wc -l ${file02} | awk '{print $1}') orphans."
echo

cp -a "${file02}" /root/orphan-results.$(date "+%Y-%m-%d.%H%M")

rm "${file01}" "${file02}"

echo "finish run @ " $(date "+%Y-%m-%d %H:%M") | tee /tmp/orphan-stoptime

By they way, you could use mktemp to avoid overwriting files in /tmp by accident, even if this is a minor improvement.

[jongeek]

Please don't pastebin such scripts, as they may get lost over time.

Thanks, I'll post any scripts directly into the thread from now on. Didn't really think about the pastebin expiration when I posted.

By they way, you could use mktemp to avoid overwriting files in /tmp by accident, even if this is a minor improvement.

Cool, updated my script and testing it out now. Thanks for the tip !

[magowiz]

Do also the .ph files doesn't need to be reported as orphaned like .pyc or they are really orphaned files if they don't belong to any package ?

[jongeek]

Depending on where they are, sure. All of the *.ph files on my machine are under /usr/lib/perl5/site_perl, and are filtered by excluding that directory. In contrast, I have *.pyc files under /usr/lib/portage and elsewhere, which is why I have a separate filter for those as well as filtering /usr/lib/python/2.5/site_packages.

[magowiz]

Depending on where they are, sure. All of the *.ph files on my machine are under /usr/lib/perl5/site_perl, and are filtered by excluding that directory. In contrast, I have *.pyc files under /usr/lib/portage and elsewhere, which is why I have a separate filter for those as well as filtering /usr/lib/python/2.5/site_packages.

thanks for the explanation.

Thank you all, it's time to add the tag solved to the title I hope someone else will find useful this topic like I do.