We have graphite tcp listeners behind an ELB in AWS (not that is is specific to AWS, just showing our scenario). When the ELB ips change, we start to not receive metrics from Graylog. This change basically just uses the proper Graphite constructor to handle dns in the graphite implementation. More info here https://github.com/dropwizard/metrics/issues/606.
As a side note, it appears at first glance that the Gelf metric reporter may suffer from this issue too. But we don't use that, so I cannot confirm that.
We have graphite tcp listeners behind an ELB in AWS (not that is is specific to AWS, just showing our scenario). When the ELB ips change, we start to not receive metrics from Graylog. This change basically just uses the proper Graphite constructor to handle dns in the graphite implementation. More info here https://github.com/dropwizard/metrics/issues/606.
As a side note, it appears at first glance that the Gelf metric reporter may suffer from this issue too. But we don't use that, so I cannot confirm that.