Is the dashboard metric average data the sum of all app instances?

1

Hi. I was wondering if the values for inbound/outbound data metrics in the fly.io dashboard represent the average network bandwidth for a single app instance or if this is the sum of all instances?

Thanks

1 Like
2

Looks like it shows both.

The prometheus datasource is per-organization. You can see the actual query in the explore section of the “Network I/O graph” (found in the drop-down menu when you click the title):

topk_avg(5,
  sum(rate(fly_instance_net_recv_bytes
   {app=~"app",device="eth0"}))by(instance,region)
, "region=Other") > 0

If you remove this query you can see that the per-instance data disappears. The allocs in those labels match what my app’s fly status does.

The same holds true if you remove the queries representing the app’s total recv/sent network bytes, ie:

sum(rate(fly_instance_net_sent_bytes{app="app",device="eth0"}))

If your question is if ‘sum’ in this context meant “over time” I did find this useful post explaining a bit about how sum works while I was digging into this.

Thanks for asking! I hope this is the answer you were looking for :slightly_smiling_face:

1 Like
3

I see! Thank you for this detailed answer! :slightly_smiling_face:

1 Like