Closed i-doit181 closed 6 years ago
This is indeed true of Rancher 1.3 as well. I'm using Telegraf and InfluxDB to access data from the Docker stats api and if your Cattle stack is managed there is no such networking metadata. I assumed this was by design. When changing the network from managed to bridged, it becomes available.
Yes but I have an use case where i cannot change network mode from managed to bridged
Yes so do I, I think most use cases based on Rancher probably require managed.
+1 having network stats provided in a standard way (via docker stats) is a real need, as it could then be used by standard tools (like Telegraf).
The rancher stats API which provides those network metrics (by the use of network namespace calls) is actually pretty hard to manipulate/use (websocket, needs to renew the token frequently, does not see new containers appearing during the current session, fixed refresh rate of 1 second, etc.) and not easily usable with existing tools.
We need a way to monitor container network usage, this is a big issue for us, would it be possible for rancher to expose these metrics in an easy format that can be read by most monitoring tools ?
Any progress on this ?
With the release of Rancher 2.0, development on v1.6 is only limited to critical bug fixes and security patches.
Rancher Version: v1.2.2 Docker Version: v1.11.2-v1.12.5 OS and where are the hosts located? (cloud, bare metal, etc): cloud with Ubuntu 15.10(4.2.0) VMs (Vagrant) with RancherOS (4.4.10) Setup Details: (single node rancher vs. HA rancher, internal DB vs. external DB) single node rancher and internal DB Environment Type: (Cattle/Kubernetes/Swarm/Mesos) Cattle Steps to Reproduce: Deploying a stack with managed network Results: docker stats shows no metrics for NET I/O for containers in managed network. A look on
docker inspect none
shows that containers in managed network are deployed with NetworkMode=none Expected: Metrics for NET I/O where traffic is going through containers