Add Munin plugin for Ceph utilization

This commit is contained in:
Joshua Boniface 2021-09-15 10:33:52 -04:00
parent be6a3992c1
commit 6d101df1ff
1 changed files with 325 additions and 0 deletions

View File

@ -0,0 +1,325 @@
#!/bin/bash
# -*- sh -*-
: << =cut
=head1 NAME
ceph_utilization - Plugin to monitor a Ceph cluster's utilization
=head1 CONFIGURATION
Defaults (no config required) for the total utilization thresholds:
[ceph_utilization]
env.warning 80
env.critical 90
=head1 AUTHOR
Joshua Boniface <joshua@boniface.me>
=head1 LICENSE
GPLv3
=head1 BUGS
=back
=head1 MAGIC MARKERS
#%# family=auto
#%# capabilities=autoconf
=cut
. "$MUNIN_LIBDIR/plugins/plugin.sh"
is_multigraph
warning=80
critical=90
RADOSDF_CMD="/usr/bin/sudo /usr/bin/rados df --format json"
OSDDF_CMD="/usr/bin/sudo /usr/bin/ceph osd df --format json"
JQ_CMD="/usr/bin/jq"
output_usage() {
echo "This plugin outputs information about a Ceph cluster"
exit 0
}
output_autoconf() {
$RADOSDF_CMD &>/dev/null
radosdf_ret=$?
$OSDDF_CMD &>/dev/null
osddf_ret=$?
$JQ_CMD --version &>/dev/null
jq_ret=$?
if [[ ${radosdf_ret} -eq 0 && ${osddf_ret} -eq 0 && ${jq_ret} -eq 0 ]]; then
echo "yes"
elif [[ ${radosdf_ret} -ne 0 || ${osddf_ret} -ne 0 ]]; then
echo "no (no 'rados' or 'ceph' command found)"
elif [[ ${jq_ret} -ne 0 ]]; then
echo "no (no 'jq' command found)"
else
echo "no (general failure)"
fi
}
output_config() {
# Graph set 1 - Ceph cluster utilization
echo 'multigraph cluster_utilization'
echo 'graph_title Cluster Utilization'
echo 'graph_args --base 1000'
echo 'graph_vlabel % Utilization'
echo 'graph_category ceph'
echo 'graph_info This graph shows the cluster utilization.'
echo 'cluster_utilization.label Cluster Utilization'
echo 'cluster_utilization.type GAUGE'
echo 'cluster_utilization.max 100'
echo 'cluster_utilization.info Percentage utilization of the cluster.'
print_warning cluster_utilization
print_critical cluster_utilization
# Graph set 2 - Ceph cluster objects
echo 'multigraph cluster_objects'
echo 'graph_title Cluster Objects'
echo 'graph_args --base 1000'
echo 'graph_vlabel Objects'
echo 'graph_category ceph'
echo 'graph_info This graph shows the cluster object count.'
echo 'cluster_objects.label Cluster Objects'
echo 'cluster_objects.type GAUGE'
echo 'cluster_objects.min 0'
echo 'cluster_objects.info Total objects in the cluster.'
POOL_LIST="$( $RADOSDF_CMD | jq -r '.pools[].name' )"
# Graph set 3 - Cluster I/O Bytes Lifetime
echo 'multigraph pool_rdbytes'
echo "graph_title IO Bytes (Lifetime)"
echo "graph_args --base 1000"
echo "graph_vlabel bytes read (-) / write (+)"
echo "graph_category ceph"
echo "graph_info This graph shows the lifetime cluster bytes."
for pool in ${POOL_LIST}; do
# Graph set 3 - Cluster I/O Bytes Lifetime
echo "pool_rdbytes_${pool}.label Pool ${pool} IO (Bytes)"
echo "pool_rdbytes_${pool}.type GAUGE"
echo "pool_rdbytes_${pool}.min 0"
echo "pool_rdbytes_${pool}.draw LINE1"
echo "pool_rdbytes_${pool}.graph no"
echo "pool_wrbytes_${pool}.label Pool ${pool} IO (Bytes)"
echo "pool_wrbytes_${pool}.type GAUGE"
echo "pool_wrbytes_${pool}.min 0"
echo "pool_wrbytes_${pool}.draw LINE1"
echo "pool_wrbytes_${pool}.negative pool_rdbytes_${pool}"
done
# Graph set 4 - Cluster I/O Operations Lifetime
echo 'multigraph pool_rdops'
echo "graph_title IO Operations (Lifetime)"
echo "graph_args --base 1000"
echo "graph_vlabel IOs read (-) / write (+)"
echo "graph_category ceph"
echo "graph_info This graph shows the lifetime cluster IOs."
for pool in ${POOL_LIST}; do
# Graph set 4 - Cluster I/O Operations Lifetime
echo "pool_rdops_${pool}.label Pool ${pool} IO (Ops)"
echo "pool_rdops_${pool}.type GAUGE"
echo "pool_rdops_${pool}.min 0"
echo "pool_rdops_${pool}.draw LINE1"
echo "pool_rdops_${pool}.graph no"
echo "pool_wrops_${pool}.label Pool ${pool} IO (Ops)"
echo "pool_wrops_${pool}.type GAUGE"
echo "pool_wrops_${pool}.min 0"
echo "pool_wrops_${pool}.draw LINE1"
echo "pool_wrops_${pool}.negative pool_rdops_${pool}"
done
# Graph set 5 - Ceph pool objects
echo 'multigraph pool_objects_total'
echo "graph_title Objects"
echo "graph_args --base 1000"
echo "graph_vlabel Objects"
echo "graph_category ceph"
echo "graph_info This graph shows the cluster object count."
for pool in ${POOL_LIST}; do
# Graph set 5 - Ceph pool objects
echo "pool_objects_total_${pool}.label Pool ${pool} Objects"
echo "pool_objects_total_${pool}.type GAUGE"
echo "pool_objects_total_${pool}.min 0"
echo "pool_objects_total_${pool}.info Total objects in the pool."
done
# Graph set 6 - Ceph pool objects copies
echo 'multigraph pool_objects_copies'
echo "graph_title Objects Copies"
echo "graph_args --base 1000"
echo "graph_vlabel Objects"
echo "graph_category ceph"
echo "graph_info This graph shows the cluster object copy count."
for pool in ${POOL_LIST}; do
# Graph set 6 - Ceph pool objects copies
echo "pool_objects_copies_${pool}.label Pool ${pool} Objects Copies"
echo "pool_objects_copies_${pool}.type GAUGE"
echo "pool_objects_copies_${pool}.min 0"
echo "pool_objects_copies_${pool}.info Total object copies in the pool."
done
# Graph set 7 - Ceph pool objects degraded
echo 'multigraph pool_objects_degraded'
echo "graph_title Objects Degraded"
echo "graph_args --base 1000"
echo "graph_vlabel Objects"
echo "graph_category ceph"
echo "graph_info This graph shows the cluster object degraded count."
for pool in ${POOL_LIST}; do
# Graph set 7 - Ceph pool objects degraded
echo "pool_objects_degraded_${pool}.label Pool ${pool} Objects Degraded"
echo "pool_objects_degraded_${pool}.type GAUGE"
echo "pool_objects_degraded_${pool}.min 0"
echo "pool_objects_degraded_${pool}.info Total degraded objects in the pool."
done
OSD_LIST="$( $OSDDF_CMD | jq -r '.nodes[].id' | sort -n )"
# Graph set 8 - Ceph OSD status
echo 'multigraph osd_status'
echo "graph_title OSD Status"
echo "graph_args --base 1000"
echo "graph_vlabel Status Up (1) / Down (0)"
echo "graph_category ceph"
echo "graph_info This graph shows the OSD status."
for osd in ${OSD_LIST}; do
# Graph set 8 - Ceph OSD status
echo "osd_status_${osd}.label osd.${osd} Status"
echo "osd_status_${osd}.type GAUGE"
echo "osd_status_${osd}.min 0"
echo "osd_status_${osd}.max 1"
echo "osd_status_${osd}.info Status of the OSD."
done
# Graph set 9 - Ceph OSD utilization
echo 'multigraph osd_utilization'
echo "graph_title OSD Utilization"
echo "graph_args --base 1000"
echo "graph_vlabel % Utilization"
echo "graph_category ceph"
echo "graph_info This graph shows the OSD utilization."
for osd in ${OSD_LIST}; do
# Graph set 9 - Ceph OSD utilization
echo "osd_utilization_${osd}.label osd.${osd} Utilization"
echo "osd_utilization_${osd}.type GAUGE"
echo "osd_utilization_${osd}.max 100"
echo "osd_utilization_${osd}.info Utilization of the OSD."
done
exit 0
}
output_values() {
RADOS_JSON_OUTPUT="$( $RADOSDF_CMD )"
OSD_JSON_OUTPUT="$( $OSDDF_CMD )"
cluster_utilization="$( $JQ_CMD -r '.total_used' <<<"${RADOS_JSON_OUTPUT}" )"
cluster_size="$( $JQ_CMD -r '.total_space' <<<"${RADOS_JSON_OUTPUT}" )"
pct_utilization="$( echo "scale=4; ${cluster_utilization} / ${cluster_size} * 100" | bc -l )"
cluster_objects="$( $JQ_CMD -r '.total_objects' <<<"${RADOS_JSON_OUTPUT}" )"
echo "multigraph cluster_utilization"
echo "cluster_utilization.value ${pct_utilization}"
echo "multigraph cluster_objects"
echo "cluster_objects.value ${cluster_objects}"
cluster_pool_count="$( $JQ_CMD -r '.pools[].name' <<<"${RADOS_JSON_OUTPUT}" | wc -l )"
echo "multigraph pool_rdbytes"
for id in $( seq 0 $(( ${cluster_pool_count} - 1 )) ); do
pool="$( $JQ_CMD -r ".pools[$id].name" <<<"${RADOS_JSON_OUTPUT}" )"
pool_rdbytes="$( $JQ_CMD -r ".pools[$id].read_bytes" <<<"${RADOS_JSON_OUTPUT}" )"
pool_wrbytes="$( $JQ_CMD -r ".pools[$id].write_bytes" <<<"${RADOS_JSON_OUTPUT}" )"
echo "pool_rdbytes_${pool}.value ${pool_rdbytes}"
echo "pool_wrbytes_${pool}.value ${pool_wrbytes}"
done
echo "multigraph pool_rdops"
for id in $( seq 0 $(( ${cluster_pool_count} - 1 )) ); do
pool="$( $JQ_CMD -r ".pools[$id].name" <<<"${RADOS_JSON_OUTPUT}" )"
pool_rdops="$( $JQ_CMD -r ".pools[$id].read_ops" <<<"${RADOS_JSON_OUTPUT}" )"
pool_wrops="$( $JQ_CMD -r ".pools[$id].write_ops" <<<"${RADOS_JSON_OUTPUT}" )"
echo "pool_rdops_${pool}.value ${pool_rdops}"
echo "pool_wrops_${pool}.value ${pool_wrops}"
done
echo "multigraph pool_objects_total"
for id in $( seq 0 $(( ${cluster_pool_count} - 1 )) ); do
pool="$( $JQ_CMD -r ".pools[$id].name" <<<"${RADOS_JSON_OUTPUT}" )"
pool_objects="$( $JQ_CMD -r ".pools[$id].num_objects" <<<"${RADOS_JSON_OUTPUT}" )"
echo "pool_objects_total_${pool}.value ${pool_objects}"
done
echo "multigraph pool_objects_copies"
for id in $( seq 0 $(( ${cluster_pool_count} - 1 )) ); do
pool="$( $JQ_CMD -r ".pools[$id].name" <<<"${RADOS_JSON_OUTPUT}" )"
pool_copies="$( $JQ_CMD -r ".pools[$id].num_object_copies" <<<"${RADOS_JSON_OUTPUT}" )"
echo "pool_objects_copies_${pool}.value ${pool_copies}"
done
echo "multigraph pool_objects_degraded"
for id in $( seq 0 $(( ${cluster_pool_count} - 1 )) ); do
pool="$( $JQ_CMD -r ".pools[$id].name" <<<"${RADOS_JSON_OUTPUT}" )"
pool_degraded="$( $JQ_CMD -r ".pools[$id].num_objects_degraded" <<<"${RADOS_JSON_OUTPUT}" )"
echo "pool_objects_degraded_${pool}.value ${pool_degraded}"
done
cluster_osd_count="$( $JQ_CMD -r '.nodes[].id' <<<"${OSD_JSON_OUTPUT}" | wc -l)"
echo "multigraph osd_status"
for id in $( seq 0 $(( ${cluster_osd_count} - 1 )) ); do
osd="$( $JQ_CMD -r ".nodes[$id].id" <<<"${OSD_JSON_OUTPUT}" )"
osd_status="$( $JQ_CMD -r ".nodes[$id].status" <<<"${OSD_JSON_OUTPUT}" )"
case ${osd_status} in
up)
osd_status="1"
;;
*)
osd_status="0"
;;
esac
echo "osd_status_${osd}.value ${osd_status}"
done
echo "multigraph osd_utilization"
for id in $( seq 0 $(( ${cluster_osd_count} - 1 )) ); do
osd="$( $JQ_CMD -r ".nodes[$id].id" <<<"${OSD_JSON_OUTPUT}" )"
osd_utilization="$( $JQ_CMD -r ".nodes[$id].utilization" <<<"${OSD_JSON_OUTPUT}" )"
echo "osd_utilization_${osd}.value ${osd_utilization}"
done
}
case $# in
0)
output_values
;;
1)
case $1 in
autoconf)
output_autoconf
;;
config)
output_config
;;
*)
output_usage
exit 1
;;
esac
;;
*)
output_usage
exit 1
esac