# Etcd keeps crashing
**URL:** https://forums.suse.com/t/etcd-keeps-crashing/10090
**Category:** Rancher 2.0 Tech Preview
**Created:** [April 16, 2018, 6:02am UTC](https://forums.suse.com/t/etcd-keeps-crashing/10090 "2018-04-16T06:02:24Z")
**Posts on this page:** 4
**Page:** 1
### Author:  [@vjtherancher](https://forums.suse.com/u/vjtherancher)
#### Post date: [April 16, 2018, 6:02am UTC](https://forums.suse.com/t/etcd-keeps-crashing/10090/1 "2018-04-16T06:02:24Z")
I am running rancher2 preview on rancheros VMs and while the cluster is coming up, the etcd container keeps crashing with this message on the UI
`[etcd] Failed to bring up Etcd Plane: Can't stop Docker container [etcd] for host [192.168.179.45]: error during connect: Post http://%2Fvar%2Frun%2Fdocker.sock/v1.24/containers/etcd/stop: tunnel disconnect`
Any idea if this is a bug or am I doing something incorrect?
Thanks.
---
### Author:  [@superseb](https://forums.suse.com/u/superseb)
#### Post date: [April 16, 2018, 7:13am UTC](https://forums.suse.com/t/etcd-keeps-crashing/10090/2 "2018-04-16T07:13:41Z")
I’ve seen this when I was running tests and I capped the network bandwidth to 100k up an down. How is the network between the host running `rancher/server` and `rancher/agent` ?
---
### Author:  [@vjtherancher](https://forums.suse.com/u/vjtherancher)
#### Post date: [April 16, 2018, 2:57pm UTC](https://forums.suse.com/t/etcd-keeps-crashing/10090/3 "2018-04-16T14:57:00Z")
Actually all vm’s are running on the same box over tap interface and the vm’s os is rancheros.
When I looked in the vm running etcd I see it does get stopped by `rancher/server`during provisioning step but doesn’t start again. I did manual restarts of etcd but the provisioning repeats and then etcd stops again with the message I mentioned in the UI.
---
### Author:  [@vjtherancher](https://forums.suse.com/u/vjtherancher)
#### Post date: [April 17, 2018, 7:00am UTC](https://forums.suse.com/t/etcd-keeps-crashing/10090/4 "2018-04-17T07:00:46Z")
With exact same setup, and the `rancher/server:stable`, the cluster comes up fine!
So it is certainly not the environment issue for sure.