AWS CloudWatch MetricMath报错排查：MetricWidget项数超限问题

我是CloudWatch MatchExpression新手，想绘制5xx错误占比。用Ruby构建了带数学表达式的Metric Widget，嵌入wiki时出现错误：MetricWidget/metrics/1 should not have more than 1 item。以下是Ruby生成的Widget代码和AWS CloudWatch控制台导出的配置，求帮忙调试。

Ruby构建的Widget代码 #

{:title=>"5xx Errors", :view=>"timeSeries", :stacked=>false, :start=>"-P7D", :period=>300, :yAxis=>{"left"=>{:min=>0, :max=>100}}, :annotations=>{:horizontal=>[{:color=>"#ff7f00", :label=>"10", :value=>10}, {:color=>"#ff0000", :label=>"50", :value=>50}]}, :metrics=>[[{"id"=>"percent_5xx_error", "expression"=>"100*(5xx/(2xx+3xx+4xx+5xx))", "label"=>"IAD", "accountId"=>"967992492170", "region"=>"us-east-1"}], ["TangerineBox", "StatusCode2xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"2xx", "stat"=>"Sum", "visible"=>false}], ["TangerineBox", "StatusCode3xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"3xx", "stat"=>"Sum", "visible"=>false}], ["TangerineBox", "StatusCode4xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"4xx", "stat"=>"Sum", "visible"=>false}], ["TangerineBox", "StatusCode5xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"5xx", "stat"=>"Sum", "visible"=>false}]]}

AWS CloudWatch控制台导出的配置 #

{
    "title": "5xx Errors",
    "view": "timeSeries",
    "stacked": false,
    "period": 300,
    "yAxis": {
        "left": {
            "min": 0,
            "max": 100
        }
    },
    "annotations": {
        "horizontal": [
            {
                "color": "#ff7f00",
                "label": "10",
                "value": 10
            },
            {
                "color": "#ff0000",
                "label": "50",
                "value": 50
            }
        ]
    },
    "metrics": [
        [ { "id": "percent_5xx_error", "expression": "100*(5xx/(2xx+3xx+4xx+5xx))", "label": "IAD", "accountId": "967992492170", "region": "us-east-1" } ],
        [ "TangerineBox", "StatusCode2xx", "ConsoleName", "fsx-console", { "accountId": "967992492170", "label": "IAD", "id": "2xx", "stat": "Sum", "visible": false } ],
        [ ".", "StatusCode3xx", ".", ".", { "accountId": "967992492170", "label": "IAD", "id": "3xx", "stat": "Sum", "visible": false } ],
        [ ".", "StatusCode4xx", ".", ".", { "accountId": "967992492170", "label": "IAD", "id": "4xx", "stat": "Sum", "visible": false } ],
        [ ".", "StatusCode5xx", ".", ".", { "accountId": "967992492170", "label": "IAD", "id": "5xx", "stat": "Sum", "visible": false } ]
    ],
    "width": 1401,
    "height": 754,
    "region": "us-east-1"
}

问题原因及修复方案 #

错误原因 #

报错指向表达式类型的metric项格式错误。CloudWatch Metric Widget规则明确：表达式metric必须是单个对象，不能嵌套在数组里。

你Ruby代码里的第一个metrics项把表达式对象包裹在了数组[ {...} ]中，这违反了格式要求，而控制台导出的嵌套数组写法是兼容格式，嵌入wiki时需要严格遵循API规范。

修复后的Ruby代码 #

将metrics数组里的第一个表达式项从嵌套数组改为直接的对象：

{:title=>"5xx Errors", :view=>"timeSeries", :stacked=>false, :start=>"-P7D", :period=>300, :yAxis=>{"left"=>{:min=>0, :max=>100}}, :annotations=>{:horizontal=>[{:color=>"#ff7f00", :label=>"10", :value=>10}, {:color=>"#ff0000", :label=>"50", :value=>50}]}, :metrics=>[
  {"id"=>"percent_5xx_error", "expression"=>"100*(5xx/(2xx+3xx+4xx+5xx))", "label"=>"IAD", "accountId"=>"967992492170", "region"=>"us-east-1"},
  ["TangerineBox", "StatusCode2xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"2xx", "stat"=>"Sum", "visible"=>false}],
  ["TangerineBox", "StatusCode3xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"3xx", "stat"=>"Sum", "visible"=>false}],
  ["TangerineBox", "StatusCode4xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"4xx", "stat"=>"Sum", "visible"=>false}],
  ["TangerineBox", "StatusCode5xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"5xx", "stat"=>"Sum", "visible"=>false}]
]}

额外优化 #

可以借鉴控制台写法，用.复用前一个metric的命名空间、维度等属性，减少重复代码：

["TangerineBox", "StatusCode2xx", "ConsoleName", "fsx-console", {"accountId"=>"967992492170", "region"=>"us-east-1", "label"=>"IAD", "id"=>"2xx", "stat"=>"Sum", "visible"=>false}],
[".", "StatusCode3xx", ".", ".", {"accountId"=>"967992492170", "label"=>"IAD", "id"=>"3xx", "stat"=>"Sum", "visible"=>false}],
[".", "StatusCode4xx", ".", ".", {"accountId"=>"967992492170", "label"=>"IAD", "id"=>"4xx", "stat"=>"Sum", "visible"=>false}],
[".", "StatusCode5xx", ".", ".", {"accountId"=>"967992492170", "label"=>"IAD", "id"=>"5xx", "stat"=>"Sum", "visible"=>false}]

内容的提问来源于stack exchange，提问作者user2284140