UBC failcnt reset
One of the frequently asked question is How do I reset failcnt in /proc/user_beancounters? While it's not suggested, it is possible.
Contents
What is failcnt?
There are a number of resource limits (called User Beancounters, or UBC for short) set for a container. If one of those resources hit its limit, the appropriate fail counter (last column of /proc/user_beancounters) increases. See Resource shortage for more info.
How to clear failcnt?
You do not need to, and this would be an incorrect thing to do.
There can be many applications that read /proc/user_beancounters, and thus if you will reset it, you may have problems with those other apps. Consider what happens if you reset your sent/received packets/bytes statistics on a network interface — programs which track it may not function properly.
Therefore, the proper usage of failcnt is not to check whether it is zero or not, but to check whether it is increased since the previous readout. In other words, check the difference, not the absolute value. There is a utility vzubc(8) which can be used for that purpose. Also, see #Bash script below.
OK, I understand, but I still want to clear failcnt!
UBC failcnts are stored for the duration of the uptime of your container. Thus, restarting the container resets the counts.
The problem here is tcp time wait buckets can still there after a container is stopped. You can check that by seeing the held column for kmemsize parameter. If it is not zero, that means you have to wait about 5 minutes in order to time wait buckets to expire, and the corresponding beancounter to be uncharged.
If you still see failcnt not reset to 0 after more than 5 minutes after container is stopped, your kernel was likely compiled with CONFIG_UBC_KEEP_UNUSED=y, and in that case you'll have to switch off this option if you want to reset beancounters when container is restarted.
If you're sure your kernel was NOT compiled with the above option and it's not resetting failcnt after 5 minutes, it means there is a bug in UBC code. Please file a detailed bug report to bugzilla.openvz.org
vzubc
vzubc(8) is a tool to show user beancounters in a decent human readable form. Its relative mode (option -r or --relative) is used to show the failcnt difference from the previous run. vzubc is available in vzctl package since vzctl-3.0.27.
Alternative bash script
This script can show the failcount deltas for one or all containers since last reset, and reset the failcounts for one or all containers.
It uses only standard commands and programs that are usually included in all linux distro's: bash, cat, grep, awk, head, tail, printf
Installation
Create the script and save it somewhere you can use it. This can either be in your home directory, or in a directory mentioned in your path variable (eg. /usr/sbin)
The script is intended to be used as root because it needs write permissions in a directory /var/lib/beanc. If that directory doesn't exist it needs write permissions in /var/lib to be able to create that directory.
You can also run it as any other user that has access to /proc/user_beancounters if you manually fix the permissions on /var/lib/beanc.
Show the beancounters
beanc show
this compares the contents of /proc/user_beancounters and the reference file, and shows you the delta value if no reference file exists, a copy of the user_beancounters file is used. to only show you the failcounts for 1 container, just add the ctid or container name to the command
beanc show mailserver beanc show 102
Reset failcounters
beanc reset mailserver beanc reset 102 beanc reset --> will reset failcounters for all containers
Confirmation will be asked.
Show only failcounts > 0
beanc brief
Initialize reference file
beanc init
This will check if the app-directory exists (/var/lib/beanc) and create it if necessary It also creates a reference file (/var/lib/beanc/user_beancounters)
BEWARE, this command will overwrite any existing reference file !
beanc source code
#!/bin/bash
#####################################################################################################################
#
# This script is intended to check the failcounts in user_beancounters on a OpenVZ machine.
# Since there is no solid way to reset the failcounts, this script maintains a copy of the file and shows delta's
#
#
# Filename: beanc
# Version : 0.1
#
#
# License:
# -------
#
# By using this script you agree there are absolutely no limitations on using it. Ofcourse there are also
# absolutely no guarantees. Please review the code to make sure it will work for you as expected.
#
# Feel free to distribute and/or modify the script.
#
# Only thing I will not appreciate is that you change my name into yours, and act like you wrote this script
# But hey, why would you do that ? And how will I ever know ?
#
# If you make changes, decide to distribute the script or feel the urge to give me feedback, please let me know.
#
#
# Author(s):
# ---------
#
# Written by Steven Broos, 7/7/2011 in a boring RHEL course
# Steven@Bit-IT.be
#
#
# Usage:
# -----
#
# 1. Copy the file to a location in your path, for example '/sbin/'
# 2. Make sure the file can be executed ('chmod 500 /sbin/beanc')
# 3. Use the script ;-) At first execution the reference file will be created in '/var/lib/beanc/'
# Note that the script is written for intended use by root, and on the OpenVZ host system
# Possibly this script wont work in a cron-job or inside a container. This has not been tested
#
#
# Options:
# -------
#
# 1. Show the beancounters
#
# beanc show
#
# this compares the contents of /proc/user_beancounters and the reference file, and shows you the delta value
# if no reference file exists, a copy of the user_beancounters file is used.
# to only show you the failcounts for 1 container, just add the ctid or container name to the command
#
# beanc show mailserver
# beanc show 102
#
# 2. To reset the failcounters for a container (or all containers):
#
# beanc reset mailserver
# beanc reset 102
# beanc reset --> will reset failcounters for all containers
#
# Confirmation will be asked.
#
# 3. It is also possible to only show the failcounters > 0
#
# beanc brief
#
# 4. If for some reason you want to manually initialize the reference file, you can execute
#
# beanc init
#
# This will check if the app-directory exists (/var/lib/beanc) and create it if necessary
# It also creates a reference file (/var/lib/beanc/user_beancounters)
# BEWARE, this command will overwrite any existing reference file !
#
#####################################################################################################################
#### ####
#### Declaration of some variables. Feel free to adjust ####
#### ####
#####################################################################################################################
rspath='/var/lib/beanc'
rsfile="$rspath/user_beancounters"
lines=24
#####################################################################################################################
#### ####
#### Function declarations. See at the bottom of the script for execution ####
#### ####
#####################################################################################################################
# show brief help message
function help ()
{
echo "$0 { reset | show | brief | init } [ <vzid> | <vzname> ]"
exit 0
}
# check existence of path and reference file, and create if necessary
# This function is executed by every start of the script
function init ()
{
if [ ! -d "$rspath" ] || [ "$1" != "" ]
then
mkdir -p "$rspath"
fi
if [ ! -f "$rsfile" ] || [ "$1" != "" ]
then
cat /proc/user_beancounters > "$rsfile"
fi
}
# Reset the failcounters by putting the current values from /proc/user_beancounters into the reference file
# either for all containers (cat > ref), or for one container (block per block)
function reset ()
{
if [ "$1" == "" ]
then
echo -n "Reset failcounts for all containers ? [y/N] "
read -n 1 yn
echo
if [ "$yn" == "y" ]
then
cat /proc/user_beancounters > "$rsfile"
fi
else
echo -n "Reset failcounts for container '`vzname $1`' ($1) ? [y/N] "
read -n 1 yn
echo
if [ "$yn" == "y" ]
then
mv "$rsfile" "${rsfile}_"
for ctid in `vzlist -Ho ctid`
do
if [ $ctid -eq $1 ]
then
echo "Resetting '`vzname $ctid`' ($ctid)"
cat /proc/user_beancounters | getblock $ctid >> "$rsfile"
else
echo "Keeping '`vzname $ctid`' ($ctid)"
cat "${rsfile}_" | getblock $ctid >> "$rsfile"
fi
done
rm -f "${rsfile}_"
fi
fi
}
# Get one or more lines from the middle of the given text : returns the 'head' of a 'tail"
# $1 : start at line
# $2 : give this much lines (optional, default 1)
# example : cat /path/file | getline 10 5
function getline ()
{
start=$1
length=$2
if [ "$length" == "" ]
then
length=1
fi
cat - | head -n $(($start+$length-1)) | tail -n $length
}
# Get all beancounter values for a container
# $1 : ctid
# example : cat /proc/user_beancounters | getblock 102
function getblock ()
{
cat - > "$rspath/tmp"
start=`cat -n "$rspath/tmp" | grep " $1:" | awk '{ print $1 }'`
echo start $start
cat "$rspath/tmp" | getline $start $lines
rm -f "$rspath/tmp"
}
# Show the contents from /proc/user_beancounters, and substitute the failcounts by a delta with the failcounts in
# the reference file
# $1 : ctid (optional, if none given all running containers are processed)
# example : show 102
function show ()
{
if [ "$1" == "" ]
then
for ctid in `vzlist -Ho ctid`
do
show $ctid
done
else
cstart=`cat /proc/user_beancounters -n | grep " $1:" | awk '{ print $1 }'`
hstart=`cat "$rsfile" -n | grep " $1:" | awk '{ print $1 }'`
for ln in `seq 0 $(($lines-1))`
do
current=`cat /proc/user_beancounters | getline $(($cstart+$ln))`
if [ "$hstart" == "" ]
then
history=''
else
history=`cat "$rsfile" | getline $(($hstart+$ln))`
fi
if [ $ln -eq 0 ]
then
resource=`echo "$current" | awk '{ print $2 }'`
held=`echo "$current" | awk '{ print $3 }'`
maxheld=`echo "$current" | awk '{ print $4 }'`
barrier=`echo "$current" | awk '{ print $5 }'`
limit=`echo "$current" | awk '{ print $6 }'`
currfcnt=`echo "$current" | awk '{ print $7 }'`
histfcnt=`echo "$history" | awk '{ print $7 }'`
fgcolor white
echo ' ----------------------------------------------------------------------------------------------------------------------
--'
printf "|%14s : %-12s %89s |\n" $1 `vzname $1` "`vzfqdn $1` (`vzip $1`)"
printf "|%14s%21s%21s%21s%21s%21s |\n" resource held maxheld barrier limit failcnt
echo ' ----------------------------------------------------------------------------------------------------------------------
--'
fgcolor reset
else
resource=`echo "$current" | awk '{ print $1 }'`
held=`echo "$current" | awk '{ print $2 }'`
maxheld=`echo "$current" | awk '{ print $3 }'`
barrier=`echo "$current" | awk '{ print $4 }'`
limit=`echo "$current" | awk '{ print $5 }'`
currfcnt=`echo "$current" | awk '{ print $6 }'`
histfcnt=`echo "$history" | awk '{ print $6 }'`
fi
if [ "$histfcnt" == "" ]
then
failcnt=$currfcnt
else
failcnt=$(($currfcnt-$histfcnt))
fi
printf ' '
if [ $failcnt -gt 0 ]
then
bgcolor red
fgcolor white
echo -en '*'
else
fgcolor green
echo -en ' '
fi
printf "%13s%21s%21s%21s%21s%21s %.s" $resource $held $maxheld $barrier $limit $failcnt "($histfcnt-$currfcnt)"
fgcolor reset
bgcolor reset
echo
done
fi
}
# Shows only counters > 0, by grepping on '|' and '-' (for the header), and '*' (for the matching failcounter)
# If no '*' has been found, nothing for that container will be shown
function showbrief ()
{
echo 'Calculating ...'
for ctid in `vzlist -Ho ctid`
do
result=`show $ctid`
matches=`echo -en "$result" | grep '*' | wc -l`
if [ $matches -gt 0 ]
then
echo -en "$result" | grep -e '|' -e '*' -e '-'
fi
done
echo 'Done!'
}
# This function returns the ctid for a container.
# $1 : container-name or even the ctid itself, since you cannot know what value the user has given
function vzid ()
{
vzlist -o name,ctid | grep -e "^$1 " -e " $1$" | awk '{ print $2 }'
}
# Returns the IP for the given ctid
# $1 : ctid
function vzip ()
{
vzlist -o ctid,ip | grep " $1 " | awk '{ print $2 }'
}
# Returns the fully qualified domain name for the given ctid
# $1 : ctid
function vzfqdn ()
{
vzlist -o ctid,hostname | grep " $1 " | awk '{ print $2 }'
}
# Returns the short name for the given ctid
# $1 : ctid
function vzname ()
{
vzlist -o ctid,name | grep " $1 " | awk '{ print $2 }'
}
# This function outputs the escape code for the specified textcolor
# $1 : Color by name (black, blue, ...)
# Or reset to revert to the terminal's default colors
# example : fgcolor red; echo 'red text'; fgcolor reset
function fgcolor ()
{
case $1 in
'black') echo -en "\033[1;30m" ;;
'green') echo -en "\033[1;32m" ;;
'red') echo -en "\033[1;31m" ;;
'cyan') echo -en "\033[1;36m" ;;
'white') echo -en "\033[1;37m" ;;
'reset') tput sgr0 ;;
esac
}
# This function outputs the escape code for the specified backgroundcolor
# $1 : Color by name (black, blue, ...)
# Or reset to revert to the terminal's default colors
# example : bgcolor green; echo 'highlighted text'; bgcolor reset
function bgcolor ()
{
case $1 in
black) echo -en "\033[0;40m" ;;
green) echo -en "\033[0;42m" ;;
red) echo -en "\033[0;41m" ;;
cyan) echo -en "\033[0;46m" ;;
white) echo -en "\033[0;47m" ;;
reset) tput sgr0 ;;
esac
}
#####################################################################################################################
#### ####
#### Execution ####
#### ####
#####################################################################################################################
# If no options are given, show a brief help message
if [ "$1" == "" ]
then
help
fi
# Check initialisation
init
# Get container ID (either by name or ID)
vzid=`vzid $2`
# If a container has been specified ($2) but none was found, give a warning and exit
if [ "$vzid" == "" ] && [ "$2" != "" ]
then
echo "Container $2 not running"
exit 1
fi
# Check what to do and call the right functions
case $1 in
'init')
init 1
;;
'show')
# If no container has been specifief, a lot of output can be expected
# Automatically send it through 'less'. option '-R' makes sure formattion will be kept
if [ "$2" == "" ]
then
show $vzid #| less -R
else
show $vzid
fi
;;
'brief')
showbrief
;;
'reset')
reset $vzid
;;
*)
help;
esacTool
See OpenVZ Bug #1298.
