diff --git a/node-daemon/monitoring/munin/pvc b/node-daemon/monitoring/munin/pvc index da3c7fa6..a8be5ccc 100755 --- a/node-daemon/monitoring/munin/pvc +++ b/node-daemon/monitoring/munin/pvc @@ -28,6 +28,8 @@ GPLv3 . "$MUNIN_LIBDIR/plugins/plugin.sh" +is_multigraph + warning=1 critical=2 @@ -36,6 +38,7 @@ PVC_CMD="/usr/bin/pvc --quiet --cluster local status --format json-pretty" JQ_CMD="/usr/bin/jq" output_usage() { + echo "This plugin outputs information about a PVC cluster and node" exit 0 } @@ -57,33 +60,58 @@ output_autoconf() { } output_config() { - echo 'graph_title PVC CHealth' - echo 'graph_args --base 100' - echo 'graph_vlabel Count' + echo 'multigraph pvc_cluster_health' + echo 'graph_title PVC Cluster Health' + echo 'graph_args --base 1000' + echo 'graph_vlabel Health%' echo 'graph_category pvc' - echo 'graph_period second' - echo 'graph_info These graphs show the health of the PVC cluster and specific node.' + echo 'graph_info Health of the PVC cluster' - echo 'pvc_cluster.label Cluster Health' - echo 'pvc_cluster.type GAUGE' - echo 'pvc_cluster.max 100' - echo 'pvc_cluster.info Health of the PVC cluster in %.' + echo 'pvc_cluster_health.label Cluster Health' + echo 'pvc_cluster_health.type GAUGE' + echo 'pvc_cluster_health.max 100' + echo 'pvc_cluster_health.min 0' + echo 'pvc_cluster_health.info Health of the PVC cluster in %' + + echo 'multigraph pvc_cluster_alert' + echo 'graph_title PVC Cluster Alerting' + echo 'graph_args --base 1000' + echo 'graph_vlabel State' + echo 'graph_category pvc' + echo 'graph_info Alerting state of the PVC cluster health' echo 'pvc_cluster_alert.label Cluster Health State' echo 'pvc_cluster_alert.type GAUGE' - echo 'pvc_cluster_alert.max 2', + echo 'pvc_cluster_alert.max 2' + echo 'pvc_cluster_alert.min 0' echo 'pvc_cluster_alert.info Alerting state of the PVC cluster health' print_warning pvc_cluster_alert print_critical pvc_cluster_alert - echo 'pvc_node.label Node Health' - echo 'pvc_node.type GAUGE' - echo 'pvc_node.max 100' - echo 'pvc_node.info Health of the PVC node in %.' + echo 'multigraph pvc_node_health' + echo 'graph_title PVC Node Health' + echo 'graph_args --base 1000' + echo 'graph_vlabel Health%' + echo 'graph_category pvc' + echo 'graph_info Health of the PVC node' + + echo 'pvc_node_health.label Node Health' + echo 'pvc_node_health.type GAUGE' + echo 'pvc_node_health.max 100' + echo 'pvc_node_health.min 0' + echo 'pvc_node_health.info Health of the PVC node in %' + + echo 'multigraph pvc_node_alert' + echo 'graph_title PVC Node Alerting' + echo 'graph_args --base 1000' + echo 'graph_vlabel State' + echo 'graph_category pvc' + echo 'graph_info Alerting state of the PVC node health' echo 'pvc_node_alert.label Node Health State' echo 'pvc_node_alert.type GAUGE' - echo 'pvc_node_alert.max 2', + echo 'pvc_node_alert.max 2' + echo 'pvc_node_alert.min 0' echo 'pvc_node_alert.info Alerting state of the PVC node health' print_warning pvc_node_alert print_critical pvc_node_alert @@ -99,8 +127,9 @@ output_values() { cluster_health="$( $JQ_CMD ".cluster_health.health" <<<"${PVC_OUTPUT}" | tr -d '"' )" cluster_health_messages="$( $JQ_CMD -r ".cluster_health.messages | @csv" <<<"${PVC_OUTPUT}" | tr -d '"' | sed 's/,/, /g' )" - echo "pvc_cluster.value ${cluster_health}" - echo "pvc_cluster.extinfo ${cluster_health_messages}" + echo 'multigraph pvc_cluster_health' + echo "pvc_cluster_health.value ${cluster_health}" + echo "pvc_cluster_health.extinfo ${cluster_health_messages}" if [[ ${cluster_health} -le 50 && ${is_maintenance} == "false" ]]; then cluster_health_alert=2 @@ -109,12 +138,14 @@ output_values() { else cluster_health_alert=0 fi + echo 'multigraph pvc_cluster_alert' echo "pvc_cluster_alert.value ${cluster_health_alert}" node_health="$( $JQ_CMD ".node_health.${HOST}.health" <<<"${PVC_OUTPUT}" | tr -d '"' )" node_health_messages="$( $JQ_CMD -r ".node_health.${HOST}.messages | @csv" <<<"${PVC_OUTPUT}" | tr -d '"' | sed 's/,/, /g' )" - echo "pvc_node.value ${node_health}" - echo "pvc_node.extinfo ${node_health_messages}" + echo 'multigraph pvc_node_health' + echo "pvc_node_health.value ${node_health}" + echo "pvc_node_health.extinfo ${node_health_messages}" if [[ ${node_health} -le 50 && ${is_maintenance} == "false" ]]; then node_health_alert=2 @@ -123,6 +154,7 @@ output_values() { else node_health_alert=0 fi + echo 'multigraph pvc_node_alert' echo "pvc_node_alert.value ${node_health_alert}" }