pvc/monitoring/munin/pvc

#!/bin/bash
# -*- sh -*-

: << =cut

=head1 NAME

pvc - Plugin to monitor a PVC cluster.

=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=0.99
critical=1.99

export PVC_CLIENT_DIR="/run/shm/munin-pvc"
PVC_CMD="/usr/bin/sudo /usr/bin/pvc --quiet cluster status --format json-pretty"
JQ_CMD="/usr/bin/jq"

output_usage() {
    echo "This plugin outputs information about a PVC cluster and node"
    exit 0
}

output_autoconf() {
    $PVC_CMD &>/dev/null
    pvc_ret=$?
    $JQ_CMD --version &>/dev/null
    jq_ret=$?

    if [[ ${pvc_ret} -eq 0 && ${jq_ret} -eq 0 ]]; then
        echo "yes"
    elif [[ ${pvc_ret} -ne 0 ]]; then
        echo "no (no 'pvc' command found or local cluster not usable)"
    elif [[ ${jq_ret} -ne 0 ]]; then
        echo "no (no 'jq' command found)"
    else
        echo "no (generic failure)"
    fi
}

output_config() {
    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_info Health of the PVC cluster'

    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.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 '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.min 0'
    echo 'pvc_node_alert.info Alerting state of the PVC node health'
    print_warning pvc_node_alert
    print_critical pvc_node_alert

    exit 0
}

output_values() {
    PVC_OUTPUT="$( $PVC_CMD )"
    HOST="$( hostname --short )"

    is_maintenance="$( $JQ_CMD ".maintenance" <<<"${PVC_OUTPUT}" | tr -d '"' )"

    cluster_health="$( $JQ_CMD ".cluster_health.health" <<<"${PVC_OUTPUT}" | tr -d '"' )"
    cluster_health_messages="$( $JQ_CMD -r ".cluster_health.messages | map(.text) | join(\", \")" <<<"${PVC_OUTPUT}" )"
    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
    elif [[ ${cluster_health} -le 90 && ${is_maintenance} == "false" ]]; then
        cluster_health_alert=1
    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 | join(\", \")" <<<"${PVC_OUTPUT}" )"
    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} != "true" ]]; then
        node_health_alert=2
    elif [[ ${node_health} -le 90 && ${is_maintenance} != "true" ]]; then
        node_health_alert=1
    else
        node_health_alert=0
    fi
    echo 'multigraph pvc_node_alert'
    echo "pvc_node_alert.value ${node_health_alert}"
}

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