Description of problem: Run quick install in env with one master specified, it will install a separate etcd instance in master instead of an embedded one and there will be [etcd] group in hosts file generated by installer tool. Version-Release number of selected component (if applicable): atomic-openshift-utils-3.3.34-1.git.0.5b998bb.el7.noarch How reproducible: always Steps to Reproduce: 1.run quick install #atomic-openshift-installer install 2.only one master specified according to hint, and the installation sumarry is as follows: *** Installation Summary *** Hosts: - openshift-118.lab.eng.nay.redhat.com - OpenShift master - OpenShift node - Etcd (Embedded) - Storage Total OpenShift masters: 1 Total OpenShift nodes: 1 Actual results: After installation completed, checked that it actually install a separated etcd but not an embedded one as described in summary. Btw, the hosts file generated by installer had a [etcd] group. [OSEv3:children] nodes masters nfs etcd Expected results: It should be an embedded etcd installed and no [etcd] in hosts. Additional info:
We actually really want to deploy external etcd rather than using embedded because this is a more robust solution.
(In reply to Scott Dodson from comment #1) > We actually really want to deploy external etcd rather than using embedded > because this is a more robust solution. @Scott If we deploy external etcd instead of embeded one, then what means "embeded" in the summary message output by install tool as following in this case: <--snip--> *** Installation Summary *** Hosts: - openshift-118.lab.eng.nay.redhat.com - OpenShift master - OpenShift node - Etcd (Embedded) - Storage Total OpenShift masters: 1 Total OpenShift nodes: 1 <--snip--> and the hint message for user: <-snip--> If only one master is specified, an etcd instance is embedded within the OpenShift master service to use as the datastore. This can be later replaced with a separate etcd instance, if required. If multiple masters are specified, then a separate etcd cluster is configured with each master serving as a member. <--snip--> The actual deploy is not corresponding with the interactive hint for user. So i think it is a bug which need to be fixed. And if we choose a more robust solution(using external etcd), then we should update the output message to get rid of misunderstanding.
Sorry, didn't realize the text said that. Assigning to Sam to get that bit updated.
(In reply to Scott Dodson from comment #3) > Sorry, didn't realize the text said that. Assigning to Sam to get that bit > updated. Nothing seriously:) I thought it is an embed one according those messages and should be what we want, so i didn't specify the misunderstanding between them again, which maybe not clear for u to catch.
Commit pushed to master at https://github.com/openshift/openshift-ansible https://github.com/openshift/openshift-ansible/commit/48861efef6dab6bde2d540ffb362f57f8650b6ba Update installation summary for etcd members We have changed how we assign etcd members in the quick installer. This updates our messaging to match those changes. Fixes Bug 1383961 Fixes Bug 1389649
Version: atomic-openshift-utils-3.4.16-1.git.0.c846018.el7.noarch Steps: 1.run quick install #atomic-openshift-installer install Result: 1. there is no embeded or separated etcd hint message in "*** Host Configuration ***" now. 2. the installation summary have been updated to be corresponding with actual deploy(separated one) *** Installation Summary *** Hosts: - openshift-118.lab.eng.nay.redhat.com - OpenShift master - OpenShift node - Etcd - openshift-116.lab.eng.nay.redhat.com - Storage 3. the install succeed and etcd has been deployed as a standalone process in master host
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2016:2778