Using aws-cloudwatch trigger with KEDA for auto scaling and not sure what the expression
attribute is! Isn't supposed to return the value of the expected metrics/dimension at polling time? If yes, why do we still need other attributes like metricCollectionTime
, metricStat
, and metricStatPeriod
?
Option 1:
Specify expression
- The docs say that it's "Optional, Required when dimensionName & dimensionValue are not specified". However, it could say namespace
, metricName
, dimensionName
, and dimensionValue
.
e.g.
...
expression: SELECT AVG("my_metric_name") FROM "Metric/Namespace" WHERE dimension1 = "dimension1value"
Option 2: namespace
, metricName
, dimensionName
, metricStat
, and dimensionValue
should be specified in place of expression
.
e.g. The equivalent to the above would be:
...
metricName: metric_name
namespace: Metric/Namespace
dimensionName: dimension1
metricStat: Average
dimensionValue: dimension1value
metricCollectionTime
and metricStatPeriod
are both needed regardless of which of the above two options is used, because they determine the time frame for which to look for metrics, whereas the above options have nothing to do with the timeframe associated to each metric.
metricCollectionTime
is how far back to look for metricsmetricStatPeriod
is how far back to aggregate metrics which are retrieved (must be less than or equal to metricCollectionTime
). The aggregation function depends on metricStat
or expression
.