Stats for host/containers always show as "Connecting..."

Hey all,

I’m experimenting with Rancher on top of a CoreOS Vagrant VM. I’m running both the server and the agent on the same machine, so I’ve followed the specific instructions for that in the docs. For some reason, everything in the interface is working apart from the stats UI (CPU, Memory, Network, Storage) on the Host and Container pages. They just keep refreshing with ‘Connecting…’ all the time.

Not sure where to start looking for issues. Any suggestions?

Not what you want to hear, but I’m having exactly the same issue. I have the Rancher host and cattle all on CoreOS in VMware guests.

Could this be because the stats rely on a service that only exists in the RancherOS implementation? If this is the case then it should be a stats-provider container that can be started on any implementation, but we’d need to know what service to add…

I might try setting up a RancherOS-based client in the cluster for comparison, but CoreOS is so much easier under VMware…

I managed to track down the issue (was obviously sleep deprived when I first investigated, because I didn’t think to just look at the logs on the Rancher agent)- seems it’s already been raised on GitHub:

The issues sounds a little like

Looking into the rancher-agent logs, I also see the fail to get systemUUID presented in the #2205 issue (linked above), but I don’t get the fatal error.

I have the same problem the host runs ubuntu 14:04 and docker from the docker repository

Version of docker Client:
Version: 1.12.3
API version: 1.24
Go version: go1.6.3
Git commit: 6b644ec
Built: Wed Oct 26 21:44:32 2016
OS/Arch: linux/amd64

Version of docker Server:
Version: 1.12.3
API version: 1.24
Go version: go1.6.3
Git commit: 6b644ec
Built: Wed Oct 26 21:44:32 2016
OS/Arch: linux/amd64

Rancher version info
Component Version
Rancher v1.1.4
Cattle v0.165.8.4
User Interface v1.1.14
Rancher Compose v0.8.6

I checked errors in the rancher-client but none are found just a number if INFO messages the same for rancher-server

I hope there is a solution for this.


Yes, in the next version of Rancher (1.2) we are moving away from using cadvisor and are using docker stats for container stats and using a packge called to collect host-level info

This occurs when running over multiple subnets, vpn or a proxy, i don’t know which. When accessing it on the local subnet it works.

I know this is an old post. I’m sorry if this fills up your messaging box.

I think I managed to “solve” part of the problem.

1 - The “client” instance you should do “Prune” from Docker. (I think it may be optional - But to know more Look for “Prune System” in the Docker documentation). Do not prune the “Rancher Server” instance.

2 - Update the instances time (clock) with your machine (the one you use to access the Rancher UI) I have this reference for this

PS. All instances need to be with time in “sync”.

3 - How you cleaned thoroughly with Docker Prune System. Raise a new Rancher client. As usual.

4 - Open “Developer Tools” in Chrome. (I think it gives different aproachs in different browsers for this process) Then you go to the “Application” tab and then in the sidebar go search again for “Application” and then you will see a “Clear Storage” option. Click “Clear Site Data”.

This way you will clear everything related to the Rancher UI. Any cookie or whatever. This also works for issues related to other services. Like Youtube, it sometimes “bug” it and gives a grotesque server error. Claiming that you did something prohibited. Just do the fourth step that solves this. You’ll just have the job of re-entering all Google accounts again. Cuz clean of all Google services.

Good luck.