install fails with oc: command not found

[Kyle-Payne]

I tried installing this a number of times tonight but I keep getting errors that it can't find the OC command. I saw other posts about this which said it was fixed in 3.11. but I'm running 3.11. I'm attaching the install text output.

I see I get this very early on.

TASK [Run variable sanity checks] *****************************************************************************************************
fatal: [10.10.10.139]: FAILED! => {"msg": "last_checked_host: 10.10.10.139, last_checked_var: openshift_master_identity_providers;Found removed variables: openshift_node_kubelet_args is replaced by openshift_node_groups[].edits; "}
to retry, use: --limit @/root/installcentos/openshift-ansible/playbooks/prerequisites.retry
install-output.txt

[albertolaley]

I have the same issue

[DarkMukke]

Since 3.10, openshift_node_kubelet_args is deprecated.
Personally i commented it out, since it will fall back to the defaults and I can always set my nodes per host later, first 2 command are to clean the server of what it tried to install (from the official docs)

Also make sure SELinux is enabled or the install will fail.

yum -y remove openshift openshift-* etcd docker
rm -rf /etc/openshift /var/lib/openshift /etc/etcd \
    /var/lib/etcd /etc/sysconfig/openshift* /etc/sysconfig/docker* \
    /root/.kube/config /etc/ansible/facts.d /usr/share/openshift
export SCRIPT_REPO=https://raw.githubusercontent.com/DarkMukke/installcentos/master
cd ~/installcentos && ./install-openshift.sh

I also found that the client is no longer shipped with the server:
openshift/origin#4773

wget -qO- https://github.com/openshift/origin/releases/download/v3.11.0/openshift-origin-client-tools-v3.11.0-0cbc58b-linux-64bit.tar.gz | tar xzv
cd openshift-origin-client-*
cp kubectl /usr/local/bin/ && cp oc /usr/local/bin/

But you might not need to install the oc manually if the installer runs completely.
After all that the install went fine

Although for whatever reasont he ansible-playbook for logging returned a non-zero, so i reran if afer:

ansible-playbook -i inventory.ini openshift-ansible/playbooks/openshift-logging/config.yml

And the second time it ran just fine. No idea what happened there

[andloh]

same issue

[faisalwaseem]

Since 3.10, openshift_node_kubelet_args is deprecated.
Personally i commented it out, since it will fall back to the defaults and I can always set my nodes per host later, first 2 command are to clean the server of what it tried to install (from the official docs)

Also make sure SELinux is enabled or the install will fail.

yum -y remove openshift openshift-* etcd docker
rm -rf /etc/openshift /var/lib/openshift /etc/etcd \
    /var/lib/etcd /etc/sysconfig/openshift* /etc/sysconfig/docker* \
    /root/.kube/config /etc/ansible/facts.d /usr/share/openshift
export SCRIPT_REPO=https://raw.githubusercontent.com/DarkMukke/installcentos/master
cd ~/installcentos && ./install-openshift.sh

I also found that the client is no longer shipped with the server:
openshift/origin#4773

wget -qO- https://github.com/openshift/origin/releases/download/v3.11.0/openshift-origin-client-tools-v3.11.0-0cbc58b-linux-64bit.tar.gz | tar xzv
cd openshift-origin-client-*
cp kubectl /usr/local/bin/ && cp oc /usr/local/bin/

But you might not need to install the oc manually if the installer runs completely.
After all that the install went fine

Although for whatever reasont he ansible-playbook for logging returned a non-zero, so i reran if afer:

ansible-playbook -i inventory.ini openshift-ansible/playbooks/openshift-logging/config.yml

And the second time it ran just fine. No idea what happened there

For some reason its sitting at docker setting for me.

Sorry figured out Docker was having issues. Sorted thanks again for the update.

[wrohandi]

@DarkMukke thanks, it works.

This issue was also confirmed on openshift-ansible project (openshift/openshift-ansible#11069).

In conclusion, what I did was:

• Start with the fresh machine
• Change the SCRIPT_REPO variable to DarkMukke's: SCRIPT_REPO=${SCRIPT_REPO:="https://raw.githubusercontent.com/DarkMukke/installcentos/master"}
• Run the installation

[be4ndr]

Since 3.10, openshift_node_kubelet_args is deprecated.
Personally i commented it out, since it will fall back to the defaults and I can always set my nodes per host later, first 2 command are to clean the server of what it tried to install (from the official docs)
Also make sure SELinux is enabled or the install will fail.

yum -y remove openshift openshift-* etcd docker
rm -rf /etc/openshift /var/lib/openshift /etc/etcd \
    /var/lib/etcd /etc/sysconfig/openshift* /etc/sysconfig/docker* \
    /root/.kube/config /etc/ansible/facts.d /usr/share/openshift
export SCRIPT_REPO=https://raw.githubusercontent.com/DarkMukke/installcentos/master
cd ~/installcentos && ./install-openshift.sh

I also found that the client is no longer shipped with the server:
openshift/origin#4773

wget -qO- https://github.com/openshift/origin/releases/download/v3.11.0/openshift-origin-client-tools-v3.11.0-0cbc58b-linux-64bit.tar.gz | tar xzv
cd openshift-origin-client-*
cp kubectl /usr/local/bin/ && cp oc /usr/local/bin/

But you might not need to install the oc manually if the installer runs completely.
After all that the install went fine
Although for whatever reasont he ansible-playbook for logging returned a non-zero, so i reran if afer:

ansible-playbook -i inventory.ini openshift-ansible/playbooks/openshift-logging/config.yml

And the second time it ran just fine. No idea what happened there

For some reason its sitting at docker setting for me.

Sorry figured out Docker was having issues. Sorted thanks again for the update.

I have the same issue.
What did you do to fix the issue with the Docker?

[faisalwaseem]

Since 3.10, openshift_node_kubelet_args is deprecated.
Personally i commented it out, since it will fall back to the defaults and I can always set my nodes per host later, first 2 command are to clean the server of what it tried to install (from the official docs)
Also make sure SELinux is enabled or the install will fail.

yum -y remove openshift openshift-* etcd docker
rm -rf /etc/openshift /var/lib/openshift /etc/etcd \
    /var/lib/etcd /etc/sysconfig/openshift* /etc/sysconfig/docker* \
    /root/.kube/config /etc/ansible/facts.d /usr/share/openshift
export SCRIPT_REPO=https://raw.githubusercontent.com/DarkMukke/installcentos/master
cd ~/installcentos && ./install-openshift.sh

I also found that the client is no longer shipped with the server:
openshift/origin#4773

wget -qO- https://github.com/openshift/origin/releases/download/v3.11.0/openshift-origin-client-tools-v3.11.0-0cbc58b-linux-64bit.tar.gz | tar xzv
cd openshift-origin-client-*
cp kubectl /usr/local/bin/ && cp oc /usr/local/bin/

But you might not need to install the oc manually if the installer runs completely.
After all that the install went fine
Although for whatever reasont he ansible-playbook for logging returned a non-zero, so i reran if afer:

ansible-playbook -i inventory.ini openshift-ansible/playbooks/openshift-logging/config.yml

And the second time it ran just fine. No idea what happened there

For some reason its sitting at docker setting for me.
Sorry figured out Docker was having issues. Sorted thanks again for the update.

I have the same issue.
What did you do to fix the issue with the Docker?

I un-installed docker. Cleaned up the directories. Rebooted OS and clean installed Docker manually to ensure it started properly. Just monitor the logs on docker startup to see if you have any wonky errors being thrown.

[marpe4k]

I had the same issue;
TASK [Run variable sanity checks]
fatal: [x.x.x.x] FAILED ...

I was able to run the installation with the informations above :-)

my solution:
forked the repo: gshipley/installcentos

in myrepo changed
inventory.ini
removed line 27 "openshift_node_kubelet_args={'pods-per-core': ['10']}"

install-openshift.sh
changed SCRIPT_REPO to "myrepo"
export SCRIPT_REPO=${SCRIPT_REPO:="https://raw.githubusercontent.com/myrepo/installcentos/master"}

made a new CentOS7 install
cloned my git repo
run the installation with
./install-openshift.sh

install is working again, hope this helps
Thanks to all :-)

[DarkMukke]

no need to fork, i already forked you can use my repo, there is also a PR open to remove the like as 10 pods per core has been the default since OKD/Openshift 3.3

[albertolaley]

Since 3.10, openshift_node_kubelet_args is deprecated.
Personally i commented it out, since it will fall back to the defaults and I can always set my nodes per host later, first 2 command are to clean the server of what it tried to install (from the official docs)

Also make sure SELinux is enabled or the install will fail.

yum -y remove openshift openshift-* etcd docker
rm -rf /etc/openshift /var/lib/openshift /etc/etcd \
    /var/lib/etcd /etc/sysconfig/openshift* /etc/sysconfig/docker* \
    /root/.kube/config /etc/ansible/facts.d /usr/share/openshift
export SCRIPT_REPO=https://raw.githubusercontent.com/DarkMukke/installcentos/master
cd ~/installcentos && ./install-openshift.sh

I also found that the client is no longer shipped with the server:
openshift/origin#4773

wget -qO- https://github.com/openshift/origin/releases/download/v3.11.0/openshift-origin-client-tools-v3.11.0-0cbc58b-linux-64bit.tar.gz | tar xzv
cd openshift-origin-client-*
cp kubectl /usr/local/bin/ && cp oc /usr/local/bin/

But you might not need to install the oc manually if the installer runs completely.
After all that the install went fine

Although for whatever reasont he ansible-playbook for logging returned a non-zero, so i reran if afer:

ansible-playbook -i inventory.ini openshift-ansible/playbooks/openshift-logging/config.yml

And the second time it ran just fine. No idea what happened there

It works.

Thanks.