Description of problem: Triger an install agaginst existed env, the detected result is a mix of installed and uninstalled env because just master and node host are checked for installed hosts while all hosts are counted when judge the cluster is an "all installed env" or "mix of installed and uninstalled env". Version-Release number of selected component (if applicable): atomic-openshift-utils-3.4.13-1.git.0.ff1d588.el7.noarch How reproducible: always Steps to Reproduce: 1.Trigger an install against existed cluster with HA env #atomic-openshift-installer install Actual results: After "gathering information from hosts", it will show an wrong detected result as following: A mix of installed and uninstalled hosts have been detected in your environment. Please make sure your environment was installed successfully before adding new nodes. Expected results: The detected result should be "All specified hosts in specified environment are installed."
Need quick installer config file and logs from the install run.
I was able to reproduce this today > *** Installation Summary *** > > Hosts: > - m01.example.com > - OpenShift master > - OpenShift node (Unscheduled) > - Etcd (Embedded) > - Storage > - n01.example.com > - OpenShift node (Dedicated) > - n02.example.com > - OpenShift node (Dedicated) > - l01.example.com > - Load Balancer (Preconfigured) > > Gathering information from hosts... > > A mix of installed and uninstalled hosts have been detected in your environment. > Please make sure your environment was installed successfully before adding new nodes. > > Do you want to (re)install the environment? To reproduce this I was required to enter a hostname/ip address for the high-availability question. I also answered 'N' for the "reference HAProxy LB" question. If I skip the high-availability question the reported bug does not surface: > *** Installation Summary *** > > Hosts: > - m01.example.com > - OpenShift master > - OpenShift node (Unscheduled) > - Etcd (Embedded) > - Storage > - n01.example.com > - OpenShift node (Dedicated) > - n02.example.com > - OpenShift node (Dedicated) > > Gathering information from hosts... > All specified hosts in specified environment are installed. > Do you want to (re)install the environment? Interesting.
(In reply to Tim Bielawa from comment #2) > I was able to reproduce this today > > > *** Installation Summary *** > > > > Hosts: > > - m01.example.com > > - OpenShift master > > - OpenShift node (Unscheduled) > > - Etcd (Embedded) > > - Storage > > - n01.example.com > > - OpenShift node (Dedicated) > > - n02.example.com > > - OpenShift node (Dedicated) > > - l01.example.com > > - Load Balancer (Preconfigured) > > > > Gathering information from hosts... > > > > A mix of installed and uninstalled hosts have been detected in your environment. > > Please make sure your environment was installed successfully before adding new nodes. > > > > Do you want to (re)install the environment? > > To reproduce this I was required to enter a hostname/ip address for the > high-availability question. I also answered 'N' for the "reference HAProxy > LB" question. > > If I skip the high-availability question the reported bug does not surface: > > > *** Installation Summary *** > > > > Hosts: > > - m01.example.com > > - OpenShift master > > - OpenShift node (Unscheduled) > > - Etcd (Embedded) > > - Storage > > - n01.example.com > > - OpenShift node (Dedicated) > > - n02.example.com > > - OpenShift node (Dedicated) > > > > Gathering information from hosts... > > All specified hosts in specified environment are installed. > > Do you want to (re)install the environment? > > Interesting. Tim, u are right. If u input LB's hostname, then there will be a lb host in config file, which will result that checked installed host number is less than total host number in config file. It seems that we just check master and node host for if it is installed host or not, but all hosts in config file will be counted when judge the cluster is an "all installed env" or "mix of installed and uninstalled env".
I added some print info for line727-line737 in cli_isntaller.py and prove it should be the cause.
Johnny Liu, a fix has been written and merged into master. The bug should be fixed now.
New PR opened which will hopefully work correctly this time. https://github.com/openshift/openshift-ansible/pull/2729 Also, you can run the atomic-openshift-installer with the '-d' option to produce a debug file in /tmp/installer.txt. When you see those 'INSTALLER_LOG.debug()' lines, that's how you get the information they are debugging. Can you please re-run the tests against the new PR? If you have mock and tito installed you can run these commands to build and install the a-o-i package locally > rm -rf /tmp/tito > tito build --test --rpm > sudo dnf erase -y "openshift-ansible*" "atomic-openshift-utils" > sudo dnf install -y /tmp/tito/noarch/*
Waiting on PR to get merged before setting back to MODIFIED.
Commit pushed to master at https://github.com/openshift/openshift-ansible https://github.com/openshift/openshift-ansible/commit/be5fbab1e3c11458f7ec979346e627923f3efe79 Merge pull request #2729 from tbielawa/BZ1390064 Bug 1390064 - [quick install] a complete installed cluster was reported as a mix of installed and uninstalled env
Version: atomic-openshift-utils-3.4.20-1.git.0.2031d1e.el7.noarch Step: 1. Install OCP 3.4 in ha env 2. Trigger an install against existed cluster and still input all host above #atomic-openshift-installer install Result: The check result hint all specified hosts in specified environment are installed.
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/RHBA-2017:0066