Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Using tmpfs for /var/{log,lock,...}
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Documentation, Tips & Tricks
View previous topic :: View next topic  
Author Message
cwng
n00b
n00b


Joined: 21 Nov 2002
Posts: 68
Location: Singapore

PostPosted: Tue Jan 25, 2005 8:56 am    Post subject: Using tmpfs for /var/{log,lock,...} Reply with quote

This is original tips/tricks post. As you can see, the code get very simplified. :) You don't even need any undo script, because now /etc/init.d/vartmp stop undo itself.

----------

Hello all,

The thread in Using tmpfs for /tmp prompted me to experiment with using tmpfs for some /var directories that has frequent disk read/write, such as /var/log. I imagine doing so will have certain advantages:

- Speed: it speeds up the logging process
- I/O: it reduces disk I/O

The following are the scripts I have conjured, to see if anyone brave enough to try them. I have been using them for a while, no problems so far.

/etc/init.d/vartmp - This is the startup script:
Code:

#!/sbin/runscript
#
# This init script mounts various /var sub-directories into RAM-based filesystem
# (i.e. tmpfs).  There are two main advantages in doing so:
#   - speed: its speeds up logging processes
#   - disk-sync: it reduces the need for constant disk syncs
#
# Pre-requisite: you need to have the following entry in your /etc/fstab
#   none    ${VARTMP_MOUNT}    tmpfs    defaults,size=256m    0 0
#
# and make sure the ${VARTMP_MOUNT} directory is created and mounted before executing
# this script

depend() {
   need localmount
   before bootmisc
}

start() {
   ebegin "Populating /mnt/vartmp "
   for d in ${VARTMP_DIRS}; do
      unpack_vartmp ${d}
   done
   eend $?
}

stop() {
   ebegin "Storing /mnt/vartmp "
   for d in ${VARTMP_DIRS}; do
      pack_vartmp ${d}
   done
   eend $?
}

# This function load previously compressed /var diretories into the tmpfs
unpack_vartmp() {
   ebegin "Unpacking /var/${1} to /mnt/vartmp/${1} "
   #
   # Check if previously stored compression of/var diretory exists
   #
   if [ ! -e /var/.vartmp.${1}.tgz ]; then
      #
      # Creates the archives for the very first time
      #
      if [ -d /var/${1} ]; then
         ewarn "Creating packed version of /var/${1}"
         cd /var && tar -czf /var/.vartmp.${1}.tgz ${1}
      else
         ewarn "No original /var/${1}"
      fi
   fi
   #
   # Check existence of ${VARTMP_MOUNT}
   #
   if [ -z ${VARTMP_MOUNT} ]; then
      mkdir /dev/shm/vartmp
      VARTMP_MOUNT=/dev/shm/vartmp
   fi
   if [ ! -d ${VARTMP_MOUNT} ]; then
      mkdir /dev/shm/vartmp
      mount none -t tmpfs -o defaults,size=256m ${VARTMP_MOUNT}
   fi
   #
   # Actual populating of previous /var contents
   #
   if [ -e /var/.vartmp.${1}.tgz ]; then
      cd ${VARTMP_MOUNT} && tar -xzf /var/.vartmp.${1}.tgz
   else
      if [ ! -d ${VARTMP_MOUNT}/${1} ]; then
         rm -f ${VARTMP_MOUNT}/${1}
         mkdir ${VARTMP_MOUNT}/${1}
         chmod 775 ${VARTMP_MOUNT}/${1}
      fi
   fi
   #
   # link /var to ${VARTMP_MOUNT}
   #
   if [ -e /var/${1} ]; then
      rm -fR /var/${1}
   fi
   ln -sf ${VARTMP_MOUNT}/${1} /var/${1}
   eend $?
}

# This function store current /var directories into a compressed archive for later use
pack_vartmp() {
   ebegin "Packing ${VARTMP_MOUNT}/${1} "
   if [ -d ${VARTMP_MOUNT}/${1} ]; then
      cd ${VARTMP_MOUNT} && tar -czf /var/.vartmp.${1}.tgz ${1}
   fi
   eend $?
}



/etc/conf.d/vartmp - This is the configuration file:
Code:

# /etc/conf.d/vartmp
#

# VARTMP_MOUNT defines where the tmpfs base directory for /var is
VARTMP_MOUNT="/mnt/vartmp"

# VARTMP_DIRS defines which of the /var sub-directories resides on the tmpfs
VARTMP_DIRS="log lock run"


Steps - Here is what I did:

First, edit /etc/fstab to contain the line:
Code:
none      ${VARTMP_MOUNT}      tmpfs      defaults,size=256m      0 0

The size I chose is 256M, its up to you to vary. You can use a higher size since tmpfs has the nice feature of not actually using any memory until it is populated.

Next, mount the tmpfs
Code:
mkdir /mnt/vartmp
mount /mnt/vartmp


Then, edit /etc/conf.d/vartmp to your liking (particularly the VARTMP_DIRS="" variable that specify which of the subdirectories to move to tmpfs. You definitely don't want to move all of them, unless you have obscene amount of RAM, like 32G :)) and run the init script:
Code:
rc-update add vartmp boot
/etc/init.d/vartmp start


For me, I will see this:
Code:
bach cwng # rc-update add vartmp boot
 * vartmp added to runlevel boot
 * Caching service dependencies...
 * rc-update complete.
bach cwng # /etc/init.d/vartmp start
 * Populating /mnt/vartmp ...
 * Unpacking /var/log to /mnt/vartmp/log ...
 * Creating packed version of /var/log                                           [ok ]
 * Unpacking /var/lock to /mnt/vartmp/lock ...
 * Creating packed version of /var/lock                                         [ ok ]
 * Unpacking /var/run to /mnt/vartmp/run ...
 * Creating packed version of /var/run                                          [ ok ]
bach cwng # ls -l /var
total 0
drwxr-xr-x   6 root root  74 Dec 21 00:21 cache
drwxr-xr-x   3 root root  16 Oct 28 14:36 db
drwxr-xr-x   2 root root  18 Jan  4 11:10 empty
drwxrwx--T   2 root gdm   18 Jan 10 22:45 gdm
drwxr-xr-x  13 root root 150 Dec 29 19:40 lib
lrwxrwxrwx   1 root root  16 Jan 25 12:12 lock -> /mnt/vartmp/lock
lrwxrwxrwx   1 root root  15 Jan 25 12:12 log -> /mnt/vartmp/log
lrwxrwxrwx   1 root root  15 Dec 20 19:06 mail -> /var/spool/mail
lrwxrwxrwx   1 root root  15 Jan 25 12:12 run -> /mnt/vartmp/run
drwxr-xr-x   8 root root  92 Jan 14 16:25 spool
drwxr-xr-x   2 root root  18 Oct 28 14:23 state
drwxrwxrwt   5 root root  93 Jan 24 15:05 tmp


Well, all seems fine. If for any reason you wish to revert back to the original setting, I have also prepared the following script vartmp-undo.sh to undo the work:

Code:
#!/bin/bash
#
# This script undo the changes caused by the switch to using tmpfs for directories in /var

if [ ! -e /etc/conf.d/vartmp ]; then
   echo "* No /etc/conf.d/vartmp!"
   exit 1
fi

source /etc/conf.d/vartmp

if [ -z ${VARTMP_MOUNT} ]; then
   VARTMP_MOUNT=/dev/shm/vartmp
fi
if [ ! -d ${VARTMP_MOUNT} ]; then
   echo "* VARTMP_MOUNT=${VARTMP_MOUNT} does not exist!"
   exit 1
fi

for d in ${VARTMP_DIRS}; do
   if [ -d ${VARTMP_MOUNT}/${d} ]; then
      if [ -e /var/${d} ]; then
         rm -f /var/${d}
      fi
      mv ${VARTMP_MOUNT}/${d} /var
   fi
done

if [ -x /etc/init.d/vartmp ]; then
   /etc/init.d/vartmp zap
fi
rm -fR /var/.vartmp.*.tgz


Just remember to do a
Code:
rc-update del vartmp

after running the undo-script.

EDIT: Split off the original material/replies. --pjp
Back to top
View user's profile Send private message
vladgrigorescu
Guru
Guru


Joined: 11 Jan 2005
Posts: 360

PostPosted: Thu Jan 05, 2006 7:10 pm    Post subject: Reply with quote

This was really helpful, thanks!
Back to top
View user's profile Send private message
Config
Retired Dev
Retired Dev


Joined: 25 May 2003
Posts: 187
Location: Zurich, Switzerland

PostPosted: Wed Jan 11, 2006 10:25 am    Post subject: Reply with quote

Just as a side-note - putting /var/log on a ramdisk is not a good idea. If something crashes, your logs won't be available on next reboot.
_________________
Config - caught by a chronic disease called tuxmania....
Back to top
View user's profile Send private message
adsmith
Veteran
Veteran


Joined: 26 Sep 2004
Posts: 1386
Location: NC, USA

PostPosted: Wed Jan 11, 2006 1:28 pm    Post subject: Reply with quote

Config wrote:
Just as a side-note - putting /var/log on a ramdisk is not a good idea. If something crashes, your logs won't be available on next reboot.


This is VERY true, but what I do is separate logs by message level:
warn and worse go to /var/log/messages (or similar), which is on HD.
but info or debug and worse and go /tmp/log/debug, which is tmpfs. This way, I can sift through all the extra info if I want, but nothing important is lost on hard reboot.

I also broadcast logs between machines on the lan, so each machine is watching the other machines' (info and worse) logs on /tmp/log/remote (tmpfs). Hence, if one machine goes down (but not all of them), I can quickly see what happened by just going to another workstation.

redundancy is good.
Back to top
View user's profile Send private message
Cazzantonio
Bodhisattva
Bodhisattva


Joined: 20 Mar 2004
Posts: 4477
Location: In giro

PostPosted: Sat Aug 18, 2007 8:38 pm    Post subject: Reply with quote

Hi. I didn't know about this tip and I created a my own initscript to to do the same things.
This script generates a tar of /var/log, /var/run and /var/lock at shutdown and populate them at reboot. You have to add it to boot runlevel.
It also does some rotating of logs so you don't have to use logrotate. You can set a maximum file size for the log tar archive and the script will create a backup (and will delete the old logs) when the size is reached.

You have to have lines like these in fstab to mount those directories in tmpfs
Code:
varlog          /var/log/               tmpfs   mode=1755,size=100m            0 0
varrun          /var/run/               tmpfs   mode=1755,size=10m             0 0
varlock         /var/lock/              tmpfs   mode=1775,size=10m             0 0


This is the initscript:
Code:
#!/sbin/runscript

depend() {
        after localmount
        before bootmisc hostname clock logger
}

#Directory for /var/log, /var/run and /var/lock backups
DATA_DIR_BACKUP=/var
#Masx size of /var/log before rotation occurs
MAX_LOG_SIZE=50

backup() {
    cd /var
    tar cvpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
    LOGSIZE=$(du -m $DATA_DIR_BACKUP/log_bk.tar|cut -f1)
    echo "Log size = $LOGSIZE M, Rotating at $MAX_LOG_SIZE"
    if [ $LOGSIZE -gt $MAX_LOG_SIZE ]; then
        echo "Log size limit reached. Rotating logs"
        /bin/rm `find /var/log -type f|grep -v "\.keep"`
        /bin/mv $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar_old -f
        tar cvzpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
    fi

    tar cvpf $DATA_DIR_BACKUP/run_bk.tar run/ &>/dev/null

    tar cvpf $DATA_DIR_BACKUP/lock_bk.tar lock/ &>/dev/null
}

restore() {
    tar xvpf $DATA_DIR_BACKUP/log_bk.tar -C /var/ &>/dev/null
    touch /var/log/{btmp,lastlog,wtmp}
    chgrp utmp /var/log/{btmp,lastlog,wtmp}
    chmod 664 /var/log/{btmp,lastlog,wtmp}

    tar xvpf $DATA_DIR_BACKUP/run_bk.tar -C /var/ &>/dev/null
    touch /var/run/utmp                             
    chgrp utmp /var/run/utmp                 
    chmod 664 /var/run/utmp

    tar xvpf $DATA_DIR_BACKUP/lock_bk.tar -C /var/ &>/dev/null
    chmod 775 /var/lock
    chown root:uucp /var/lock
}

start() {
        ebegin "Restoring /var/log, /var/run and /var/lock"
        restore
        eend $?
}

stop() {
        ebegin "Backup structure of /var/log, /var/run and /var/lock"
        backup
        eend $?
}

_________________
Any man's death diminishes me, because I am involved in mankind, and therefore never send to know for whom the bell tolls; it tolls for thee
-John Donne
Back to top
View user's profile Send private message
mv
Advocate
Advocate


Joined: 20 Apr 2005
Posts: 4308

PostPosted: Sun Aug 26, 2007 7:45 am    Post subject: Reply with quote

Cazzantonio wrote:
Hi. I didn't know about this tip and I created a my own initscript to to do the same things.

Is there a particular reason for these chown and chrgrp commands? Aren't they automatically executed by tar?
Moreover, why is it necessary to touch utmp? (Does something go crazy if utmp has "old" timestamp?)
Back to top
View user's profile Send private message
Cazzantonio
Bodhisattva
Bodhisattva


Joined: 20 Mar 2004
Posts: 4477
Location: In giro

PostPosted: Thu Oct 30, 2008 11:21 am    Post subject: Reply with quote

mv wrote:
Cazzantonio wrote:
Hi. I didn't know about this tip and I created a my own initscript to to do the same things.

Is there a particular reason for these chown and chrgrp commands? Aren't they automatically executed by tar?
Moreover, why is it necessary to touch utmp? (Does something go crazy if utmp has "old" timestamp?)
Just to make sure those files are there... They must exists for linux to behave correctly. The rest of the tar archive could become corrupted and all other files get lost, but at least you want to boot and get a working console.
_________________
Any man's death diminishes me, because I am involved in mankind, and therefore never send to know for whom the bell tolls; it tolls for thee
-John Donne
Back to top
View user's profile Send private message
Cazzantonio
Bodhisattva
Bodhisattva


Joined: 20 Mar 2004
Posts: 4477
Location: In giro

PostPosted: Thu Jan 20, 2011 8:16 pm    Post subject: Reply with quote

Hi all. I stumbled on this old script of mine and I gave it an update. Now it's divided into an init script and a log daemon. The name of the files is important, so if you want to change it you have to change also something in the two scripts.
It does a backup of the directory structure of /var/log, /var/run, /var/lock and /var/spool, restoring necessary files and subdirectories each reboot and providing a rotation of logs in case the tar file becomes too big (it preserves emerge.log).

It is compatible with baselayout-2 and openrc, at least up to version 0.7.0. It should behave also with parallel boot, but don't take it for granted...

Basically you have to put this into fstab:
Code:
varlog      /var/log/       tmpfs   mode=1755,size=100m            0 0
varrun      /var/run/      tmpfs   mode=1755,size=1m            0 0
varlock      /var/lock/      tmpfs   mode=1775,size=1m            0 0
varspool          /var/spool   tmpfs   mode=1750,size=2m            0 0


put this initscript in the boot runlevel:

/etc/init.d/manage_log :
Code:
#!/sbin/runscript

depend() {
        after localmount root
        before bootmisc logger cron
}

#Directory for backups of /var/log, /var/run and /var/lock
DATA_DIR_BACKUP=/var
#Maximum size of logs in the tar archive before rotating
MAX_LOG_SIZE=50
#If you want the script to act like a daemon and rotate logs before the system restart, then set this variable to 1
DEMONIZED=0
#If DEMONIZED=1, then select the time interval between checks (s,m,h,d are the suffixes for seconds, minutes, hours, days)
CHECK_TIME=2h
#The daemon is the script located in /usr/local/sbin/log_daemon.sh

backup() {
    cd /var
    tar cvpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
    chmod 600 $DATA_DIR_BACKUP/log_bk.tar
    LOGSIZE=$(du -m $DATA_DIR_BACKUP/log_bk.tar|cut -f1)
    if [ $LOGSIZE -gt $MAX_LOG_SIZE ]; then
        cd /var
        echo "Log size limit reached. Rotating logs"
        /bin/mv $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar_old -f
        tar cvzpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
        chmod 600 $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar
        /bin/mv $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar_old -f
        /bin/rm `find /var/log -type f|grep -v "\.keep"|grep -v "emerge.log"`
        touch /var/run/utmp; chgrp utmp /var/run/utmp; chmod 664 /var/run/utmp
        tar cvzpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
        chmod 600 $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar
    else
        echo "Log size = $LOGSIZE M, Rotating at $MAX_LOG_SIZE"
    fi
    /bin/rm `find /var/run -type f -iname *.pid` -rf
    echo "" > /var/run/utmp
    tar cvpf $DATA_DIR_BACKUP/run_bk.tar run/ &>/dev/null
    chmod 600 $DATA_DIR_BACKUP/run_bk.tar
    tar cvpf $DATA_DIR_BACKUP/lock_bk.tar lock/ &>/dev/null
    chmod 600 $DATA_DIR_BACKUP/lock_bk.tar
    tar cvpf $DATA_DIR_BACKUP/spool_bk.tar spool/ &>/dev/null
    chmod 600 $DATA_DIR_BACKUP/spool_bk.tar
}

restore() {
    tar xvpf $DATA_DIR_BACKUP/log_bk.tar -C /var/ &>/dev/null
    tar xvpf $DATA_DIR_BACKUP/run_bk.tar -C /var/ &>/dev/null
    test -e /var/run/utmp || touch /var/run/utmp && chgrp utmp /var/run/utmp && chmod 664 /var/run/utmp
    tar xvpf $DATA_DIR_BACKUP/lock_bk.tar -C /var/ &>/dev/null
    chmod 775 /var/lock && chown root:wheel /var/lock
    tar xvpf $DATA_DIR_BACKUP/spool_bk.tar -C /var/ &>/dev/null
    chmod 755 /var/spool
    if [ $DEMONIZED -eq 1 ]; then
        /usr/local/sbin/log_daemon.sh &
    fi
}

start() {
        ebegin "Recover structure of /var/log, /var/run, /var/lock and /var/spool"
        restore
        eend $? 
}

stop() {
        ebegin "Backup structure of /var/log, /var/run, /var/lock and /var/spool"
        backup
        eend $?
}


and this is the daemon in case you want to daemonize the rotation of the logs:

/usr/local/sbin/log_daemon.sh :
Code:
#!/bin/sh

DEMONIZED=`cat /etc/init.d/manage_log|grep -v "#" |grep DEMONIZED|cut -d"=" -f2|head -n1`
DATA_DIR_BACKUP=`cat /etc/init.d/manage_log|grep -v "#" |grep DATA_DIR_BACKUP|cut -d"=" -f2|head -n1`
MAX_LOG_SIZE=`cat /etc/init.d/manage_log|grep -v "#" |grep MAX_LOG_SIZE|cut -d"=" -f2|head -n1`
CHECK_TIME=`cat /etc/init.d/manage_log|grep -v "#" |grep CHECK_TIME|cut -d"=" -f2|head -n1`

while [ $DEMONIZED -eq 1 ]; do
    LOGSIZE=$(du -ms /var/log|cut -f1)
    if [ $LOGSIZE -gt $MAX_LOG_SIZE ]; then
        cd /var
        echo "Log size limit reached. Rotating logs"
        /etc/init.d/syslog-ng stop
        /bin/mv $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar_old -f
        tar cvzpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
        chmod 600 $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar
        /bin/mv $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar_old -f
        /bin/rm `find /var/log -type f|grep -v "\.keep"|grep -v "emerge.log"`
        touch /var/run/utmp; chgrp utmp /var/run/utmp; chmod 664 /var/run/utmp
        tar cvzpf $DATA_DIR_BACKUP/log_bk.tar log/ &>/dev/null
        chmod 600 $DATA_DIR_BACKUP/log_bk.tar $DATA_DIR_BACKUP/log_bk.tar
        /etc/init.d/syslog-ng start
    fi
    sleep $CHECK_TIME
done

_________________
Any man's death diminishes me, because I am involved in mankind, and therefore never send to know for whom the bell tolls; it tolls for thee
-John Donne
Back to top
View user's profile Send private message
slick
Bodhisattva
Bodhisattva


Joined: 20 Apr 2003
Posts: 3487

PostPosted: Wed Mar 23, 2011 1:59 pm    Post subject: Reply with quote

I found some apps they create its own directory in /var/log or /var/run (during install?) . If mount /var/log or /var/run with tmpfs this apps will not start because missing directories. Can anybody reproduce this? This will avoid to mount this directories empty with tmpfs in /etc/fstab, without backup and restore the structure?
Back to top
View user's profile Send private message
tclover
Guru
Guru


Joined: 10 Apr 2011
Posts: 300

PostPosted: Tue May 10, 2011 11:39 am    Post subject: Yet another vartmp script Reply with quote

Here will follow my flavour of the above scripts, the only extra script is a little 'pack' script which is called by crontab to pack in a regular basis in order to minimize data loss.
Quote:
# /etc/conf.d/vartmp
VTMP_MNT="/var/tmp/vartmp"
VTMP_DIR="mail"
VTMP_SIZE=128m

Quote:
#!/sbin/runscript
# Copyright 1999-2011 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: /etc/init.d/vartmp rev.3 $
#
# This init script mounts various /var sub-directories into RAM-based filesystem
# (i.e. tmpfs). There are two main advantages in doing so:
# - speed: its speeds up logging processes
# - disk-sync: it reduces the need for constant disk syncs
#
# Now this line, or somothing like that, with the desired size value:
# vartmp $VTMP_MNT tmpfs defaults,size=128m 0 0
# is not a mandatory anymore. vartmp will check if the required tmpfs is mounted, and if not, mount it.
# This was necessary as I runned this script for the first time and noticed nothing in the output of mount.

depend() {
need localmount
before bootmisc acpid
}

start() {
check_vartmp || return 1
[ -n "${VTMP_DIR}" ] && {
for d in ${VTMP_DIR}; do check_archive ${d}
unpack_vartmp ${d}; done; }
eend $?
}

stop() {
ebegin "storing ${VTMP_MNT}"
for d in ${VTMP_DIR}; do pack_vartmp ${d}; done
eend $?
}

check_vartmp() {
[ -z "${VTMP_MNT}" ] && VTMP_MNT=/var/vartmp
[ -z "${VTMP_SIZE}" ] && VTMP_SIZE=128m
[ -n "`mount -t tmpfs|grep ${VTMP_MNT}`" ] || { mkdir -p ${VTMP_MNT}
mount vartmp ${VTMP_MNT} -t tmpfs -o defaults,size=${VTMP_SIZE}; }
eend $?
}

check_archive() {
[ -e /var/.vartmp.${1}.tgz ] || [ -d /var/${1} ] && {
ewarn "creating packed version of /var/${1} for the first time."
cd /var && tar -czf .vartmp.${1}.tgz ${1}
eend $?
rm -rf ${1} && ln -sf ${VTMP_MNT}/${1} ${1}
eend $?; }
}

unpack_vartmp() {
ebegin "unpacking /var/${1} to ${VTMP_MNT}/${1}"
cd ${VTMP_MNT} && tar -xzf /var/.vartmp.${1}.tgz || {
mkdir -p ${VTMP_MNT}/${1} && chmod 775 ${VRTMP_MNT}/${1}; }
eend $?
}

pack_vartmp() {
ebegin "Packing ${VTMP_MNT}/${1}"
cd ${VTMP_MNT} && tar -czf /var/.vartmp.${1}.tgz ${1}
eend $?
}

Quote:
#!/bin/sh
#~/vtmp-pack.sh

. /etc/conf.d/vartmp

pack_vartmp() {
cd ${VTMP_MNT} && tar -czf /var/.vartmp.${1}.tgz ${1}
[ $? ] || echo "$?"
}

[ -n "`/etc/init.d/vartmp status|grep started`" ] && \
pack_vartmp $1 || echo "vartmp not started" && exit 1

You have to add something like this: */20 * * * * root $HOME/vtmp-pack.sh log
in crontab for this to work.
Back to top
View user's profile Send private message
d-fens
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 88

PostPosted: Fri Feb 08, 2013 10:48 am    Post subject: Reply with quote

hi,

i used cwng's script to put samba files into tmpfs, so the hdds do spin down on my nas, i had do change a bit in order to allow also nested directories to work, here is my diff:

Code:
47c47,49
<          cd /var && tar -czf /var/.vartmp.${1}.tgz ${1}
---
>          path=$1
>        escaped=${path/\//\\}
>        cd /var && tar -czf /var/.vartmp.${escaped}.tgz ${1}
66,67c68,69
<    if [ -e /var/.vartmp.${1}.tgz ]; then
<       cd ${VARTMP_MOUNT} && tar -xzf /var/.vartmp.${1}.tgz
---
>    if [ -e /var/.vartmp.${escaped}.tgz ]; then
>       cd ${VARTMP_MOUNT} && tar -xzf /var/.vartmp.${escaped}.tgz
89c91,93
<       cd ${VARTMP_MOUNT} && tar -czf /var/.vartmp.${1}.tgz ${1}
---
>       path=$1
>       escaped=${path/\//\\}
>       cd ${VARTMP_MOUNT} && tar -czf /var/.vartmp.${escaped}.tgz ${1}


so now you can use
Code:

VARTMP_DIRS="log/samba cache/samba lib/samba"


in /etc/conf.d/vartmp
Back to top
View user's profile Send private message
seneca
n00b
n00b


Joined: 10 Apr 2013
Posts: 53

PostPosted: Wed Apr 17, 2013 7:45 am    Post subject: Reply with quote

quick question. With the changes in udev >=200, do these tmpfs references have to be changed to devtmpfs?
Back to top
View user's profile Send private message
khayyam
Advocate
Advocate


Joined: 07 Jun 2012
Posts: 2247

PostPosted: Wed Apr 17, 2013 11:33 am    Post subject: Reply with quote

seneca wrote:
quick question. With the changes in udev >=200, do these tmpfs references have to be changed to devtmpfs?

seneca ... no, devtempfs is a temporary filesystem for device nodes, so /dev, tmpfs is a more general temporary filesystem, the former hasn't replaced the latter.

Code:
# grep TMPFS /usr/src/linux/.config
CONFIG_DEVTMPFS=y
CONFIG_DEVTMPFS_MOUNT=y
CONFIG_TMPFS=y
CONFIG_TMPFS_POSIX_ACL=y
CONFIG_TMPFS_XATTR=y

best ... khay
Back to top
View user's profile Send private message
seneca
n00b
n00b


Joined: 10 Apr 2013
Posts: 53

PostPosted: Wed Apr 17, 2013 11:38 am    Post subject: Reply with quote

Thank you, Khay

Oh, new to this... is it ok to thank people for their help, as now, or does this clutter the forums? will refrain in the future based on your advice
Back to top
View user's profile Send private message
khayyam
Advocate
Advocate


Joined: 07 Jun 2012
Posts: 2247

PostPosted: Wed Apr 17, 2013 12:50 pm    Post subject: Reply with quote

seneca wrote:
Thank you, Khay

seneca ... you're welcome :)

seneca wrote:
Oh, new to this... is it ok to thank people for their help, as now, or does this clutter the forums? will refrain in the future based on your advice

I generally think its good form, if it solves the issue then its some clue that the provided information worked, though tagging the thread [SOLVED] is also fine. I most often refrain from replying with a "you're welcome" unless there is something further to add :)

best ... khay
Back to top
View user's profile Send private message
ppurka
Advocate
Advocate


Joined: 26 Dec 2004
Posts: 3209

PostPosted: Wed Apr 17, 2013 3:10 pm    Post subject: Reply with quote

seneca wrote:
Oh, new to this... is it ok to thank people for their help, as now, or does this clutter the forums? will refrain in the future based on your advice
Even better. Quote the correct reply and say "thank you" :) Then some future wanderer will stumble upon the thread and know exactly which advice worked.
_________________
emerge --quiet redefined | E17 vids: I, II | Now using kde4 :-/
Back to top
View user's profile Send private message
roscop
n00b
n00b


Joined: 17 Dec 2013
Posts: 1

PostPosted: Tue Dec 17, 2013 12:39 pm    Post subject: Reply with quote

Hi,

Does anyone else had to modify initscript to handle run and lock changes or have I missed something in my last install ?

https://bugs.gentoo.org/show_bug.cgi?id=342049
https://bugs.gentoo.org/show_bug.cgi?id=332633

Regards
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Documentation, Tips & Tricks All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum