You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have identified an issue with Librato if I attempt to submit a metric where a composite metric by the same name already exists. This results in Librato returning a HTTP 400 error due to the name collision. statsd-librato identifies the non-HTTP-200 response and prints an error:
statsd 2015/07/03 09:05:02 unable to submit measurements: 400 Bad Request: {"errors":{"params":{"type":["'metric.restapi.requesttime' is a GAUGE, but was submitted as different type"]}},"request_time":1435878302}
Rather than statsd-librato moving on from the error, it appears to constantly attempt resubmission of the invalid data at each interval. New stats submissions don't occur as the erroneous data is constantly resubmitted until it is accepted - which of course it never will.
Temporary errors like HTTP 500/503..etc, should be reattempted but permanent errors such as HTTP 4xx should be logged, droped and we should move on.
The text was updated successfully, but these errors were encountered:
I have identified an issue with Librato if I attempt to submit a metric where a composite metric by the same name already exists. This results in Librato returning a HTTP 400 error due to the name collision.
statsd-librato
identifies the non-HTTP-200 response and prints an error:Rather than
statsd-librato
moving on from the error, it appears to constantly attempt resubmission of the invalid data at each interval. New stats submissions don't occur as the erroneous data is constantly resubmitted until it is accepted - which of course it never will.Temporary errors like HTTP 500/503..etc, should be reattempted but permanent errors such as HTTP 4xx should be logged, droped and we should move on.
The text was updated successfully, but these errors were encountered: