Hey there!
We are slurping up metrics via a datadog agent and curious if there is a way to emit or tag metrics with process group?
What I am trying to do is break out instance metrics by app_name and ideally process_group.
Not sure if this is in the prometheus dataset and I just need to coerce datadog to add those tags?
Appreciate the help!
A few months ago we would have said no, but now there actually is a way to group by process group that we’ve added recently. (It’s still entirely undocumented, and we haven’t updated any of the Fly Metrics dashboards to take advantage of this, and it’s a little tricky to use…)
There’s a metric fly_instance_info that includes a process_group label and the fixed value 1, which you can join against other metrics using a group modifier (e.g., group_left(process_group)) on a binary operator *.
So, for example, this query will include process_group in cpu metrics:
fly_instance_cpu * on(instance) group_left(process_group) fly_instance_info
Hope this is helpful!
Awesome. I will look into how datadog grabs metrics. On first glance it looks like it might be able to use fly_process_group from the machine api to tag those metrics.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.