Description of problem: docker exec into the etcd pod fails: docker exec -ti cc1195d1292c bash rpc error: code = 2 desc = oci runtime error: exec failed: container_linux.go:247: starting container process caused "process_linux.go:110: decoding init error from pipe caused \"read parent: connection reset by peer\"" this is causing the OKD master to kill the etcd pod repeatedly. see detailed debugging from https://github.com/openshift/origin/issues/21609 CC vrutkovs Version-Release number of selected component (if applicable): OKD 3.11 oc version oc v3.11.0+62803d0-1 kubernetes v1.11.0+d4cacc0 features: Basic-Auth GSSAPI Kerberos SPNEGO Server https://os-int.tremolo.local:443 openshift v3.11.0+06cfa24-67 kubernetes v1.11.0+d4cacc0 Docker 1.13.1 docker version Client: Version: 1.13.1 API version: 1.26 Package version: docker-1.13.1-84.git07f3374.el7.centos.x86_64 Go version: go1.10.2 Git commit: 07f3374/1.13.1 Built: Fri Nov 30 02:48:45 2018 OS/Arch: linux/amd64 Server: Version: 1.13.1 API version: 1.26 (minimum version 1.12) Package version: docker-1.13.1-84.git07f3374.el7.centos.x86_64 Go version: go1.10.2 Git commit: 07f3374/1.13.1 Built: Fri Nov 30 02:48:45 2018 OS/Arch: linux/amd64 Experimental: false CentOS 7.6 cat /etc/redhat-release CentOS Linux release 7.6.1810 (Core) How reproducible: Very, multiple users reporting the issue Steps to Reproduce: 1. Install OKD on CentOS 7.6 2. 3. Actual results: Expected results: Additional info:
Doesn't seem to be happening on OKD CI with docker-1.13.1-84.git07f3374.el7.centos.x86_64
Note, that OKD CI (openshift-ansible's PRs in particular) are using CentOS 7.5 as a base and only install docker
ran "sudo yum downgrade docker-1.13.1-75.git8633870.el7.centos.x86_64 docker-client-1.13.1-75.git8633870.el7.centos.x86_64 docker-common-1.13.1-75.git8633870.el7.centos.x86_64" on each host and was able to get back to normal
*** This bug has been marked as a duplicate of bug 1655214 ***