Why gorouter generates massive metrics
Article ID: 405439
Updated On:
Products
Issue/Introduction
On a platform with multiple gorouter and many active apps, gorouter generates very large number of logs + metrics (gauge and counter).
For example, in a medium scale platform. output of `cf log-meta` indicate that gorouters generate 1M line of logs and metric in 35 seconds.
$ cf log-meta
Retrieving log cache metadata as admin...
Source Source Type Count Expired Cache Duration
...
gorouter platform 1000000 562733143 35s
...
This could cause significant loss rate on reverse_log_proxy.
Cause
By default, gorouter emits per-request metrics, on a platform with many active apps, massive metrics can be generated and emitted per second.
Resolution
If the per-request metrics from gorouter is not required and it causes significantly high loss rate on reverse_log_proxy, you can disable or partly disable the feature at,
- Ops Manager > Tanzu Platform for Cloud Foundry > Setting > Networking > The per-request metrics that the Gorouter emits.
by changing the configuration from "Report all per-request metrics" to "Report some per-request metrics" and choose some of the available options based on need. Choosing all options is identical to "Report all per-request metrics".
- Latency and route lookup time metrics
- HttpStartStop server metrics
- HttpStartStop client metrics
Feedback
