Cloudwatch Alarm for Glue Job Failures

0

Hi all, is there any direct way to set up cloudwatch alarms to alert you when a glue job fails? Without using the lambda function. For example, using a direct metric such as glue.driver.aggregate.numFailedTasks

Thanks!

2 Answers
0
Accepted Answer

Do you have any idea if the glue.error.ALL observability group metric would work on CloudWatch alarms instead of the glue.driver.aggregate.numFailedTasks?

==>> Yes, replicated at my end with the below CloudWatch alarm configurations and its working as expected.:

Statistic : SUM

Period : 60

MetricName: glue.error.ALL

TreatMissingData :notBreaching

ComparisonOperator : GreaterThanOrEqualToThreshold

Threshold: 1

Note: So for every 1 minute if there is a data point breaching the threshold you will be getting an alert for the same.

Reference:

[1] Using Amazon CloudWatch alarms - https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/AlarmThatSendsEmail.html

[2] Create a CPU usage alarm - Setting up a CPU usage alarm using the AWS Management Console - https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/US_AlarmAtThresholdEC2.html#cpu-usage-alarm-console

AWS
SUPPORT ENGINEER
answered 5 months ago
profile picture
EXPERT
reviewed a month ago
  • Thanks! I checked the same as well and it is working as expected in sending me an email for a glue job failure.

0
  • If you want to notification on a Glue Job Failure, then you can use AWS Event Bridge, no Lambda Required.

  • Create an Event Bridge Rules like below pattern for a job failure and set the Target to be SNS Topic.

  • Subscript the SNS Topic to get a notification.

  • Sample Event Rule pattern to monitor a Glue job failure: { "detail-type": ["Glue Job State Change"], "source": ["aws.glue"], "detail": { "jobName": ["<Glue job Name>"], "state": ["FAILED"] } }

  • Check the below Article for the same: https://repost.aws/knowledge-center/glue-sns-notification-state

  • Also, using glue.driver.aggregate.numFailedTasks metrics for job notification would NOT be the correct Glue job status. As this metrics is related to Spark task, which sometime fails and succeed when spark retries. So the Glue job will succeed even with failed Spark task.

AWS
answered 5 months ago
profile picture
EXPERT
reviewed 5 months ago
  • got it. thanks so much I will have to test this out on monday as i dont have the permissions required to create a rule on eventbridge. Do you have any idea if the glue.error.ALL observability group metric would work on cloudwatch alarms instead of the glue.driver.aggregate.numFailedTasks?

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions