There is an equivalent setting in DynatraceConfig and I think that is useful. It adds a tag to all metrics with the given deviceId. If set in the Prometheus Registry, the scrape method could automatically append the deviceId tag.
@pjfanning You can easily do this with a common tag for Prometheus. deviceId is a specific required field of dynatrace that has no corollated requirement in Prometheus. Several other variants of this exist in other monitoring systems, e.g. host. I think they are all conceptually just tags.
@jkschneider I can understand that when I create a custom metric, I can add my own host/deviceId tag. I still think it would be useful for the metrics in DefaultExports (JVM, GC metrics, etc.) to be able to get a host/deviceId tag when the prometheus data is exported.
In a previous job, the prometheus exporter that we used there automatically added some tags like the host which I thought was very useful.
@pjfanning When @jkschneider mentioned 'common tags', he was referring to the common tags concept.
That will allow you to set what the deviceId is that is relevant for your use case. For example a Kubernetes deployment would care about a podId while a raw AWS deploy might care about something different.
Most helpful comment
@pjfanning When @jkschneider mentioned 'common tags', he was referring to the common tags concept.
That will allow you to set what the deviceId is that is relevant for your use case. For example a Kubernetes deployment would care about a podId while a raw AWS deploy might care about something different.