Linux Bonding Health

模板定义

template linux.bonding {
    subject = {{.Last.Status}}: {{.Eval .Alert.Vars.by_host}} bad bond(s) on {{.Group.host}}
    body = `{{template "header" .}}
    <h2>Bond Status</h2>
    <table>
    <tr><th>Bond</th><th>Slave</th><th>Status</th></tr>
    {{range $r := .EvalAll .Alert.Vars.slave_status}}
        {{if eq $.Group.host .Group.host}}
            <tr>
                <td>{{$r.Group.bond}}</td>
                <td>{{$r.Group.slave}}</td>
                <td {{if lt $r.Value 1.0}} style="color: red;" {{end}}>{{$r.Value}}</td>
            </tr>
        {{end}}
    {{end}}
    </table>
    `
}

警报定义

alert linux.bonding {
    template = linux.bonding
    macro = host_based
    $notes = This alert triggers when a bond only has a single interface, or the status of a slave in the bond is not up
    $slave_status = max(q("sum:linux.net.bond.slave.is_up{bond=*,host=*,slave=*}", "5m", ""))
    $slave_status_by_bond = sum(t($slave_status, "host,bond"))
    $slave_count = max(q("sum:linux.net.bond.slave.count{bond=*,host=*}", "5m", ""))
    $no_good = $slave_status_by_bond < $slave_count || $slave_count < 2
    $by_host = max(t($no_good, "host"))
    warn = $by_host
} 

通知 Priview

http://i.stack.imgur.com/8kolx.jpg