How to update the field via Painless Scripting in a Watcher rule

alerting

#1

We are using Watcher to monitor the CPU utilization and send the alert email to the support team.

We found that the data format of CPU usage field is a decimal and not the percent, so we try to use Painless script to convert the format of this field. However, the updated Watcher rule can't work as expected. The error message in the elasticsearch log as below:

[2016-12-09T08:34:08,077][ERROR][o.e.x.w.a.e.ExecutableEmailAction] [elk5-es-poc-node-3] failed to execute action [Linux_High_CPU_Alert/send_email]. failed to transform payload. ScriptException[runtime error]; nested: NullPointerException;

So can anyone help us to check what's wrong in the below Watcher code? Many thanks for your help.

  "trigger": {
    "schedule": {
      "interval": "1m"
    }
  },
  "input": {
    "search": {
      "request": {
        "indices": [ "<metricbeat-{now/d}>" ],
        "body": {
          "size": 0,
          "query": {
            "bool": {
              "filter": [
                {
                  "range": {
                    "@timestamp": {
                      "gt":"now-15m","lt":"now"
                    }
                  }
                },
                {
                  "range": {
                    "system.cpu.user.pct":{"gt":"0.2","lt":"2"}
                  }
                }
              ]
            }
          },
          "aggs": {
            "group_by_hostname": {
              "terms": {
                "field": "beat.hostname.keyword",
                "size": 5
              },
                "aggs": {
                    "get_latest": {
                      "terms": {
                        "field": "@timestamp",
                        "size": 1,
                        "order": {
                          "_term": "desc"
                        }
                      },
			    "aggs": {
                    "range": {
                      "date_range": {
                        "field": "@timestamp",
                        "format": "MM/dd/yyyy",
                        "ranges": [
                          { "to": "now" },
						  { "from": "now-10m" }
                        ]
					  },
                      "aggs": {
                        "group_by_cpu_pct": {
                          "terms": {
                            "field": "system.cpu.user.pct"
                          }
                        }
                      }
					}
				  }	
                }
              }
            }
          }
        }
      }
    }
  },
  "condition": {
    "compare": {
      "ctx.payload.hits.total": {
        "gt": 0
      }
    }
  },
  "actions": {
    "send_email": {
	  "transform": {
        "script": {
		  "lang": "painless",
          "inline": "ctx.payload.hits._source.system.cpu.user.pct = ctx.payload.hits._source.system.cpu.user.pct * params.percentage",
		  "params": {
             "percentage": 10
           }
        }
      },
      "email": {
        "to": [ "vinson@test.com" ], 
        "subject": "Watcher Notification - Found Server CPU High Utilization",
        "body": {
          "html": "<body><br><b>---------------------------High CPU Usage Alert---------------------------</b></br><br></br><br>The below Hosts have the high CPU usage. Please perform the OS checking accordingly!</br><br></br><br></br><table ><thead><tr><th>|</th><th>Hostname</th><th>|</th><th>Timestamp</th><th>|</th><th>CPU_Usage</th><th>|</th></tr></thead><tbody>{{#ctx.payload.aggregations.group_by_hostname.buckets}}<tr><td>|</td><td>{{key}}</td><td>| </td><td>{{#get_latest.buckets}}{{range.buckets.0.key}}</td><td>| </td><td>{{range.buckets.0.group_by_cpu_pct.buckets.0.key}}{{/get_latest.buckets}}</td><td>|</td></tr>{{/ctx.payload.aggregations.group_by_hostname.buckets}}</tbody></table><br><i>Report generated by ELK. This is system generated email, please do not reply.</i></br></body>"
        }
      }
    }
  }
}'

(Alexander Reelsen) #2

Hey,

please take the time to properly use markdown formatting and put your watch in three backticks so it gets formatted like this

I am formatted
for multi line
stuff

So, you are trying to access a field, that does not exist and this causes a NullPointerException. The field ctx.payload.hits.source is unlikely to exist, as the hits field is an array.

If you want to loop through all of the hits, something like

ctx.payload.hits.hits.forEach(hit -> hit._source.foo = hit._source.foo * 10) ; return ctx.payload

--Alex


#3

Hi Alex,

Many thanks for your reply. As you advised, I modified the Watcher as below:

  "actions": {
    "send_email": {
	    "transform": {
        "script": {
		  "lang": "painless",
          "inline": "ctx.payload.hits.hits.forEach(hit -> hit._source.system.cpu.user.pct = hit._source.system.cpu.user.pct * 100) ; return ctx.payload"
        }
      },
	    "email": {
        "to": [ "vinson@test.com" ], 
        "subject": "Watcher Notification - Found Server CPU High Utilization",
        "body": {
          "html": "<body><br><b>---------------------------High CPU Usage Alert---------------------------</b></br><br></br><br>The below Hosts have the high CPU usage. Please perform the OS checking accordingly!</br><br></br><br></br><table ><thead><tr><th>|</th><th>Hostname</th><th>|</th><th>Timestamp</th><th>|</th><th>CPU_Usage</th><th>|</th></tr></thead><tbody>{{#ctx.payload.aggregations.group_by_hostname.buckets}}<tr><td>|</td><td>{{key}}</td><td>| </td><td>{{#get_latest.buckets}}{{range.buckets.0.key}}</td><td>| </td><td>{{range.buckets.0.group_by_cpu_pct.buckets.0.key}}{{/get_latest.buckets}}</td><td>|</td></tr>{{/ctx.payload.aggregations.group_by_hostname.buckets}}</tbody></table><br><i>Report generated by ELK. This is system generated email, please do not reply.</i></br></body>"
        }
      }
    }
  }

However, I found that the format of CPU usage was still the decimal in the alert email as following. So can you please advise what's wrong in my Watcher code? Thanks.

---------------------------High CPU Usage Alert---------------------------

The below Hosts have the high CPU usage. Please perform the OS checking accordingly!
| Hostname | Timestamp | CPU_Usage |
| dbvdev4133 | *-12/14/2016 | 0.4296000003814697 |
| uklpaurzr21a | *-12/14/2016 | 0.9817000031471252 |
| uklvadbpm05a | *-12/14/2016 | 0.2985999882221222 |
| uklvadrel001a | *-12/14/2016 | 0.25040000677108765 |
| uklvadshp03a | *-12/14/2016 | 0.3197999894618988 |

Report generated by ELK. This is system generated email, please do not reply.


(Alexander Reelsen) #4

Hey,

you could use something like

def df = new DecimalFormat('##.##');
ctx.payload.hits.hits.forEach(hit -> hit._source.system.cpu.user.pct = df.format(hit._source.system.cpu.user.pct * 100)) ; return ctx.payload

--Alex


(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.