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
As mentioned in the previous section, returning a lot of metrics is a computationally intensive process.
91
+
92
+
Default endpoints `/metrics` and `/metrics/per-object` expose every possible, including low-level stats like Erlang VM stats and higher-level things like queue stats.
93
+
94
+
Using aggregation is one way to reduce the number of metrics and somewhat reduce CPU usage.
95
+
96
+
It's also possible to completely disable some groups of higher-level metrics that provide excessive level of detail, or pose no interest at all in given circumstances. For queue metrics it's also possible to filter on a per-vhost basis - that can be useful if there is a way to choose less interesting queues (like one-off transient queues for RPC) and don't include them in the output.
97
+
98
+
Those customizations can be applied to default endpoints using a configuration file. The following config snippet will make default endpoints to expose just enough information to get a number of messages and and a number of consumers for each queue, but only in the default vhost `/`:
Even when the number of metrics for the default endpoints is reduced, it's possible to get any combination of those configurable metrics via a separate endpoint, where HTTP `GET`-parameters determine what exactly should be returned. E.g. scraping `/metrics/core?vhost=vhost-1&vhost=vhost-2&family=queue_coarse_metrics&family=queue_consumer_count`. will only return requested metrics (and not, for example, low-level Erlang VM metrics). It supports the following parameters:
107
+
108
+
* Zero or more `family` - if given, only these metric families will be returned. The full list is documented in [metrics](metrics.md), and it's the same names that are being used in the config file.
109
+
* Zero or more `vhost` - if it's given, queue related metrics (`queue_coarse_metrics`, `queue_coarse_metrics` and `queue_metrics`) will be returned only for given vhost(s).
110
+
* Optional `per-object` - if it's `1` or `true`, per-object values without aggregation will be returned.
0 commit comments