Hello,
I'm trying to analyze metrics collect by "AWS ELB Application Load Balancer by HTTP" (7.0-3). In particular TargetGroups metrics.
I think it is the same for all AWS Cloudwatch API call and I think it could have an issue with the way it is done:
1. we ask to CloudWatch a list of metric for 1h start_time = end_time - AWS.request_period * 60; (AWS.request_period = 60)
2. each metric are requested every 1m (AWS.request_period)
3. we shoud receive an array of 60 values but it is not the case. if there is no data, cloudwatch api doesn't send values.
4. we take the most recent value (.Values.first().first()) but in case there was no data collected we take a previous value.
In current situation I could have a metric that stay the same during mostly one hour although it is not the good one.
Metrics are requested every 5min (m/5) so I'm wondering why we are requesting 1h of data ?
Why not request 5min with a period of 5m to have only one value ? or at least 5min with a period of 1min ?
Is it possible to adjust timestamp of a metric collected through an itemprotitype ? Because we have that value in the json metric file.
I am right in the analyze?
attached a result of a Target Group metric to illustrate. Some metrics always have 60 points but not the HTTPCode_Target_?XX_Count.
Regards
I'm trying to analyze metrics collect by "AWS ELB Application Load Balancer by HTTP" (7.0-3). In particular TargetGroups metrics.
I think it is the same for all AWS Cloudwatch API call and I think it could have an issue with the way it is done:
1. we ask to CloudWatch a list of metric for 1h start_time = end_time - AWS.request_period * 60; (AWS.request_period = 60)
2. each metric are requested every 1m (AWS.request_period)
3. we shoud receive an array of 60 values but it is not the case. if there is no data, cloudwatch api doesn't send values.
4. we take the most recent value (.Values.first().first()) but in case there was no data collected we take a previous value.
In current situation I could have a metric that stay the same during mostly one hour although it is not the good one.
Metrics are requested every 5min (m/5) so I'm wondering why we are requesting 1h of data ?
Why not request 5min with a period of 5m to have only one value ? or at least 5min with a period of 1min ?
Is it possible to adjust timestamp of a metric collected through an itemprotitype ? Because we have that value in the json metric file.
I am right in the analyze?
attached a result of a Target Group metric to illustrate. Some metrics always have 60 points but not the HTTPCode_Target_?XX_Count.
Regards