cilium / tetragon

eBPF-based Security Observability and Runtime Enforcement

Home Page:https://tetragon.io

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

sensor manager: allow concurrent listing with loading.

kkourt opened this issue · comments

5ce0f71 introduced a 0.9sec timeout to list the tracing policies from the sensor manager. The intention was to not block the metrics update for a long time.

However, users seem to be hitting the following warning in production, probably due to the sensor manager loading sensors that take a long time.

fix "error listing tracing policies to collect policies state" error="context deadline exceeded" warning messages

A solution would be to allow the listing of tracing policies to happen concurrently with loading. This can either be done via introducing a rw-lock or a new goroutine just for the loading.

Having statistics about how long a policy did take to load would also be useful.