- Repository
- Munin (2.0)
- Last change
- 2019-02-08
- Graph Categories
- Family
- auto
- Capabilities
- Keywords
- Language
- Bash
- License
- GPL-2.0-only
- Authors
vserver_resources
Name
vserver_resources - Plugin to graph Vserver resource usage and limits
Configuration
This plugin uses the following configuration variables
[vserver_resources]
env.vservers - specify the vservers to include in the graph
env.resource - specify the resource to be monitored
env.limits - if true, turn on limit graphing
Default Configuration
This is the default configuration
[vserver_resources]
env.vservers <all vservers>
env.limits false
Example Configurations
The following monitors the RSS value for the vservers named “vserver1 vserver2 vserver3 vserver4” and looks to see if the resource limit has been breached, if so it sends a message to nagios via send_nsca, and sends an email to notify that this has happened:
[vserver_resources]
user root
env.vservers vserver1 vserver2 vserver3 vserver4
env.resource RSS
env.limits 1
contacts nagios email
contact.nagios.command /usr/bin/send_nsca -H your.nagios-host.here -c /etc/send_nsca.cfg
contact.email.command mail -s "Munin-notification for ${var:group} :: ${var:host}" your@email.address.here
This second example monitors the VM value for all vservers on the system and has no limit notifications turned on:
[vserver_resources]
user root
env.vservers vserver5 vserver6 vserver7
env.resource VM
env.limits 0
This last example monitors all the resources for vserver5. Note that this will be a busy graph, and it would be really useless if you specified more than one vserver when the resource is set to ALL:
[vserver_resources]
user root
env.vservers vserver5
env.resource ALL
env.limits 0
Usage
Possible values for env.resource are:
ALL
all the below resources
PROC
number of processes
VM
sum of all virtual pages inside the guest
VML
sum of all virtual pages locked into memory
RSS
number of pages currently present in RAM
ANON
number of anonymous memory pages
FILES
number of open files
OFD
LOCKS
SOCK
MSGQ
SHM
number of shared memory pages
Authors
Copyright (C) 2006-2008 Holger Levsen, Micah Anderson
License
GNU GPLv2
Todo
Less shell subprocesses
The plugin uses lots and lots of shell subprocesses. Maybe factor common vserver plugin code into Munin::Plugin::Linux::Vserver?
Multigraph
make it so you can specify more than one resource to be graphed? Alternatively define combined ressource-display: VM+RSS+ANON+SHM and FILES+OFD+LOCK+SOCK (for one vserver only), and/or make it so you can graph all resources for one vserver
Default resource
set a default for the resource if it is unset?
Use /proc less often
use /proc less often (100 times more overhead than talking to the kernel directly) i.e. use something like pagesize=\`perl -MPOSIX -e ‘print POSIX::sysconf(_SC_PAGESIZE), “\n”;'\`
Bugs
The “ALL” resource is broken
Magic Markers
#%# family=auto
#%# capabilities=autoconf
#!@@BASH@@
# -*- sh -*-
: <<'=cut'
=head1 NAME
vserver_resources - Plugin to graph Vserver resource usage and limits
=head1 CONFIGURATION
This plugin uses the following configuration variables
[vserver_resources]
env.vservers - specify the vservers to include in the graph
env.resource - specify the resource to be monitored
env.limits - if true, turn on limit graphing
=head2 DEFAULT CONFIGURATION
This is the default configuration
[vserver_resources]
env.vservers <all vservers>
env.limits false
=head1 EXAMPLE CONFIGURATIONS
The following monitors the RSS value for the vservers named "vserver1
vserver2 vserver3 vserver4" and looks to see if the resource limit has
been breached, if so it sends a message to nagios via send_nsca, and
sends an email to notify that this has happened:
[vserver_resources]
user root
env.vservers vserver1 vserver2 vserver3 vserver4
env.resource RSS
env.limits 1
contacts nagios email
contact.nagios.command /usr/bin/send_nsca -H your.nagios-host.here -c /etc/send_nsca.cfg
contact.email.command mail -s "Munin-notification for ${var:group} :: ${var:host}" your@email.address.here
This second example monitors the VM value for all vservers on the system and
has no limit notifications turned on:
[vserver_resources]
user root
env.vservers vserver5 vserver6 vserver7
env.resource VM
env.limits 0
This last example monitors all the resources for vserver5. Note that
this will be a busy graph, and it would be really useless if you
specified more than one vserver when the resource is set to ALL:
[vserver_resources]
user root
env.vservers vserver5
env.resource ALL
env.limits 0
=head1 USAGE
Possible values for env.resource are:
=over 4
=item ALL
all the below resources
=item PROC
number of processes
=item VM
sum of all virtual pages inside the guest
=item VML
sum of all virtual pages locked into memory
=item RSS
number of pages currently present in RAM
=item ANON
number of anonymous memory pages
=item FILES
number of open files
=item OFD
=item LOCKS
=item SOCK
=item MSGQ
=item SHM
number of shared memory pages
=back
=head1 AUTHORS
Copyright (C) 2006-2008 Holger Levsen, Micah Anderson
=head1 LICENSE
GNU GPLv2
=begin comment
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 dated June, 1991.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
=end comment
=head1 TODO
=over 4
=item Less shell subprocesses
The plugin uses lots and lots of shell subprocesses. Maybe factor
common vserver plugin code into Munin::Plugin::Linux::Vserver?
=item Multigraph
make it so you can specify more than one resource to be graphed?
Alternatively define combined ressource-display: VM+RSS+ANON+SHM and
FILES+OFD+LOCK+SOCK (for one vserver only), and/or make it so you can
graph all resources for one vserver
=item Default resource
set a default for the resource if it is unset?
=item Use /proc less often
use /proc less often (100 times more overhead than talking to the
kernel directly) i.e. use something like pagesize=\`perl -MPOSIX -e
'print POSIX::sysconf(_SC_PAGESIZE), "\n";'\`
=back
=head1 BUGS
The "ALL" resource is broken
=head1 MAGIC MARKERS
#%# family=auto
#%# capabilities=autoconf
=cut
if [ "$1" = "autoconf" ]; then
if [ -r /proc/virtual/info ]; then
echo yes
else
echo "no (/proc/virtual/info not found)"
fi
exit 0
fi
VSERVERS="${vservers:-}"
LIMITS="${limits:-}"
RESOURCE="${resource:-}"
# shellcheck disable=SC2207
INFO=($(sed 's/.*:\t//' /proc/virtual/info 2>/dev/null || echo '<none>'))
KCIN=$(( 16#${INFO[2]} ))
# If this is 1, then VCI_SPACES is present in the kernel (new in 2.6.19)
if [ $(( (KCIN >> 10) & 1 )) -eq 1 ]
then
NAMELOC="nsproxy"
else
NAMELOC="cvirt"
fi
if [ -z "$VSERVERS" ] ; then
XIDS=$(find /proc/virtual/* -type d -exec basename {} \;)
else
# it's really more performant to specify vservers by ids or not at all
XIDS=""
for i in $VSERVERS ; do
if [ -d "/proc/virtual/$i" ] ; then
XIDS="${XIDS}${i} "
else
# shellcheck disable=SC2044
for j in $(find /proc/virtual/* -type d -exec basename {} \;) ; do
if [ "$i" = "$(grep "NodeName" "/proc/virtual/$j/$NAMELOC" | cut -f 2)" ]; then
XIDS="${XIDS}${j} "
fi
done
fi
done
fi
if [ "$1" = "config" ]; then
case "$RESOURCE" in
PROC)
echo 'graph_title Processes used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Processes'
echo 'graph_info Shows the number of processes used by each vserver.'
;;
VM)
echo 'graph_title Virtual memory used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel VM pages'
echo 'graph_info Shows virtual memory (human readable) used by each vserver.'
;;
VML)
echo 'graph_title Locked memory used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel VML pages'
echo 'graph_info Shows locked memory (human readable) used by each vserver.'
;;
RSS)
echo 'graph_title Resident set size used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel RSS pages'
echo 'graph_info Shows resident set size (human readable) used by each vserver.'
;;
ANON)
echo 'graph_title Anonymous memory used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel ANON pages'
echo 'graph_info Shows anonymous memory (human readable) used by each vserver.'
;;
FILES)
echo 'graph_title Files used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Files'
echo 'graph_info Shows files used by each vserver.'
;;
OFD)
echo 'graph_title Open filedescriptors used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Open filedescriptors'
echo 'graph_info Shows open filedescriptors used by each vserver.'
;;
LOCKS)
echo 'graph_title Locks used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Locks'
echo 'graph_info Shows locks used by each vserver.'
;;
SOCK)
echo 'graph_title Sockets used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Sockets'
echo 'graph_info Shows sockets used by each vserver.'
;;
MSGQ)
echo 'graph_title Message queues used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel Message queues'
echo 'graph_info Shows message queues used by each vserver.'
;;
SHM)
echo 'graph_title Shared memory used by vserver'
echo 'graph_args --base 1024k -l 0'
echo 'graph_vlabel SHM pages'
echo 'graph_info Shows shared memory (human readable) used by each vserver.'
;;
*)
echo "$RESOURCE not defined."
exit 1
;;
esac
echo 'graph_category vserver'
# do not assume we are on i386 where pagesize is 4096...
pagesize=$(perl -MPOSIX -e 'print POSIX::sysconf(_SC_PAGESIZE), "\n";')
for xid in $XIDS ; do
LABEL=$(grep "NodeName" "/proc/virtual/$xid/$NAMELOC" | cut -f 2)
NAME=$(echo "$LABEL" | cut -d. -f1 | tr '-' '_')
case "$RESOURCE" in
PROC)
echo "$NAME.label $LABEL: processes"
echo "$NAME.info Number of processes used by $LABEL."
;;
VM)
echo "$NAME.label $LABEL: Virtual memory"
echo "$NAME.info Size of virtual memory used by $LABEL. (Number multipled by $pagesize to make it human readable)"
echo "$NAME.cdef $NAME,$pagesize,*"
;;
VML)
echo "$NAME.label $LABEL: Locked memory"
echo "$NAME.info Size of locked memory used by $LABEL. (Number multipled by $pagesize to make it human readable)"
echo "$NAME.cdef $NAME,$pagesize,*"
;;
RSS)
echo "$NAME.label $LABEL: Resident set size"
echo "$NAME.info Size of resident set size used by $LABEL. (Number multiplied by $pagesize to make it human readable)"
echo "$NAME.cdef $NAME,$pagesize,*"
;;
ANON)
echo "$NAME.label $LABEL: Anonymous memory"
echo "$NAME.info Size of anonymous memory used by $LABEL. (Number multiplied by $pagesize to make it human readable)"
echo "$NAME.cdef $NAME,$pagesize,*"
;;
FILES)
echo "$NAME.label $LABEL: Files"
echo "$NAME.info Number of files used by $LABEL."
;;
OFD)
echo "$NAME.label $LABEL: Open filedescriptors"
echo "$NAME.info Number of open filedescriptors used by $LABEL."
;;
LOCKS)
echo "$NAME.label $LABEL: Locks"
echo "$NAME.info Number of locks used by $LABEL."
;;
SOCK)
echo "$NAME.label $LABEL: Sockets"
echo "$NAME.info Number of sockets used by $LABEL."
;;
MSGQ)
echo "$NAME.label $LABEL: Message queues"
echo "$NAME.info Number of message queues used by $LABEL."
;;
SHM)
echo "$NAME.label $LABEL: Shared memory"
echo "$NAME.info Size of shared memory used by $LABEL. (Number multiplied by $pagesize to make it human readable)"
echo "$NAME.cdef $1,$pagesize,*"
;;
*)
echo "$RESOURCE not defined."
exit 1
;;
esac
if [ -n "$LIMITS" ] && [ "$LIMITS" = 1 ]; then
LIMIT=$(grep "$RESOURCE" "/proc/virtual/$xid/limit" | cut -f 4)
if [ "${LIMIT:-0}" -gt 0 ]; then
echo "$NAME.critical $LIMIT"
fi
fi
done
exit 0
fi
for xid in $XIDS ; do
LABEL=$(grep "NodeName" "/proc/virtual/$xid/$NAMELOC" | cut -f 2)
NAME=$(echo "$LABEL" | cut -d. -f1 | tr '-' '_')
awk -v name="${NAME}" -v resource="${RESOURCE}:" \
'{ if ( $1 == resource ) printf "%s.value %d\n", name, $2 }' \
"/proc/virtual/$xid/limit"
done