Attempt to start Rancher on Windows2019 Server triggers "panic"


I have attempted to install/start Rancher v2.2.8 on Windows 2019 Server that has a Docker v19.03.2 with LCOW (linux containers on windows) support enabled, but getting following error (see) below) and container continues to loop endlessly, but fails to start. Looked on the documentation site, but I can’t figure out if my configuration even supported or it is just me being a newb with this. Does any one know ?

Regards, Sender.

2019/10/02 17:18:58 [INFO] Rancher version v2.2.8 is starting
2019/10/02 17:18:58 [INFO] Rancher arguments {ACMEDomains:[] AddLocal:auto Embedded:false KubeConfig: HTTPListenPort:80 HTTPSListenPort:443 K8sMode:auto Debug:false NoCACerts:false ListenConfig: AuditLogPath:/var/log/auditlog/rancher-api-audit.log AuditLogMaxage:10 AuditLogMaxsize:100 AuditLogMaxbackup:10 AuditLevel:0}
2019/10/02 17:18:58 [INFO] Listening on /tmp/log.sock
2019/10/02 17:18:58 [INFO] Running etcd --data-dir=management-state/etcd
2019-10-02 17:18:58.635745 W | pkg/flags: unrecognized environment variable ETCD_UNSUPPORTED_ARCH=amd64
2019-10-02 17:18:58.635988 I | etcdmain: etcd Version: 3.2.13
2019-10-02 17:18:58.635998 I | etcdmain: Git SHA: Not provided (use ./build instead of go build)
2019-10-02 17:18:58.636004 I | etcdmain: Go Version: go1.12.9
2019-10-02 17:18:58.636009 I | etcdmain: Go OS/Arch: linux/amd64
2019-10-02 17:18:58.636014 I | etcdmain: setting maximum number of CPUs to 2, total number of available CPUs is 2
2019-10-02 17:18:58.641188 N | etcdmain: the server is already initialized as member before, starting as etcd member…
2019-10-02 17:18:58.882098 I | embed: listening for peers on http://localhost:2380
2019-10-02 17:18:59.132435 I | embed: listening for client requests on localhost:2379
2019-10-02 17:18:59.196606 C | mvcc/backend: cannot open database at management-state/etcd/member/snap/db (invalid argument)
panic: cannot open database at management-state/etcd/member/snap/db (invalid argument)

Have you tried it with Docker 18.09? Curious if that is the issue

No I haven’t. The folks who own this environment said they had to bump the version due to some LCOW related issues and are not planning to go back.

Any ideas on how should I go about troubleshooting this would be really appreciated.

For a record, thanks to Adrian Goins, who gave me the answer through his youtube channel ( as following:

Rancher itself has to run on Linux. In 2.3 (GA on Oct 8) you can deploy Windows clusters using Windows Server 2019 with the KB4489899 hotfix applied, but you can only use Windows as the worker nodes. You’ll still need Linux servers for the control plane and one worker to handle ingress. See for the prerequisites and step-by-step guide for making it work.

1 Like