luisarizmendi/ocp-libvirt-ipi-role

Ansible Role to deploy OpenShift 4 on a CentOS 7 or Fedora KVM using libvirt IPI

ocp-libvirt-ipi-role's Issues

NFS Provisioner broken in 4.8

As an FYI the NFS Provisioner doesn't work in OCP 4.8 due to the removal of the selfLink feature from Kubernetes. Looking at switching to the NFS SubDirectory provisioner and will submit a PR when done.

https://kubernetes-sigs.github.io/nfs-subdir-external-provisioner/

Is it possible to get Terraform + Openshift/libvirt to create VMs with hugepages?

Hi Luis,

On my hypervisors, I have most of the memory dedicated to hugepages (the default 2Mb hugepages).
This allows me to fit more VMs into reserved memory and makes their TLB's smaller.

Here's an example:

# virsh dumpxml sat6|head -24
<domain type='kvm' id='3'>
  <name>sat6</name>
[...]
  <memoryBacking>
    <hugepages/>
  </memoryBacking>
[...]

I've done some research and it appears that Terraform has support for hugepages but I wouldn't know where to change it in the openshift installer.. Do you have any idea of where that might be?

On systems with a limited rootfs, the role fills up /root and the rootfs.

Hi,

First of all, thank you for creating this role. I'll have a PR soonish for some RHEL7 specifics.

on a system where I have /, /var, /usr, /home and others as dedicated logical volumes, I found the role was filling up '/' because of /root/go, /root/ocp and /root/.cache/go-build.

Is there a way to redirect the workdir to something else than the $HOME of the root user?

[root@myhost ~]# df -h /
Filesystem                  Size  Used Avail Use% Mounted on
/dev/mapper/rootdg-lv_root  3.9G  3.7G     0 100% /

[root@myhost ~]# du -ks go  .cache/ocp .cache/go*
378928  go
0       .cache/ocp
0       .cache/go
2021744 .cache/go-build

Role should use the openshift built in libvirt support

The OpenShift baremetal IPI installer has builtin support for libvirt (see https://docs.openshift.com/container-platform/4.10/installing/installing_bare_metal_ipi/ipi-install-installation-workflow.html#retrieving-the-openshift-installer_ipi-install-installation-workflow)
This libvirt support is used to create the bootstrap vm when doing the install.
Using this support would quite reduce the size of the taks in the role, and make it even more readable.

Errors when the variable lb (haproxy) is true (CentOS 8)

I used ocp-libvirt-ipi-role in my project which provision various OpenShift development environments on GCP such as:
CRC, SNC (Single Node Cluster) and MNC.

When the variable lb (haproxy) is true, I needed to make some modifications before starting ansible:

a) pip3 install netaddr

Otherwise, ansible returns an error:

...
TASK [luisarizmendi.ocp_libvirt_ipi_role : Create backends haproxy file] *******
task path: /root/.ansible/roles/luisarizmendi.ocp_libvirt_ipi_role/tasks/kvm_deploy.yml:279
fatal: [localhost]: FAILED! => {"changed": false, "msg": "AnsibleFilterError: The ipmath filter requires python's netaddr be installed on the ansible controller"}

b)

sed -i 's/mode tcp/mode http/g' .ansible/roles/luisarizmendi.ocp_libvirt_ipi_role/templates/haproxy.cfg.back.j2

Otherwise, ansible/haproxy returns an error:

...
TASK [luisarizmendi.ocp_libvirt_ipi_role : Starting services] ***************************************************************
task path: /root/.ansible/roles/luisarizmendi.ocp_libvirt_ipi_role/tasks/kvm_deploy.yml:306
failed: [localhost] (item=haproxy) => {"ansible_loop_var": "item", "changed": false, "item": "haproxy", "msg": "Unable to start service haproxy: Job for haproxy.service failed because the control process exited with error code.\nSee \"systemctl status haproxy.service\" and \"journalctl -xe\" for details.\n"}

PLAY RECAP ******************************************************************************************************************
localhost                  : ok=26   changed=9    unreachable=0    failed=1    skipped=6    rescued=0    ignored=0

...

$ sudo systemctl status haproxy.service -l --no-pager

● haproxy.service - HAProxy Load Balancer
   Loaded: loaded (/usr/lib/systemd/system/haproxy.service; enabled; vendor preset: disabled)
   Active: failed (Result: exit-code) since Fri 2021-08-13 16:41:45 UTC; 2min 36s ago
  Process: 57604 ExecStartPre=/usr/sbin/haproxy -f $CONFIG -c -q $OPTIONS (code=exited, status=1/FAILURE)

Aug 13 16:41:45 crc-build-1 systemd[1]: Starting HAProxy Load Balancer...
Aug 13 16:41:45 crc-build-1 haproxy[57604]: [ALERT] 224/164145 (57604) : http backend 'app' (/etc/haproxy/haproxy.cfg:85) tries to use incompatible tcp backend 'mycluster-openshift-api-server' (/etc/haproxy/haproxy.cfg:97) in a 'use_backend' rule (see 'mode').
Aug 13 16:41:45 crc-build-1 haproxy[57604]: [ALERT] 224/164145 (57604) : http backend 'app' (/etc/haproxy/haproxy.cfg:85) tries to use incompatible tcp backend 'mycluster-ingress-https' (/etc/haproxy/haproxy.cfg:113) in a 'use_backend' rule (see 'mode').
Aug 13 16:41:45 crc-build-1 haproxy[57604]: [ALERT] 224/164145 (57604) : http backend 'app' (/etc/haproxy/haproxy.cfg:85) tries to use incompatible tcp backend 'mycluster-ingress-http' (/etc/haproxy/haproxy.cfg:106) in a 'use_backend' rule (see 'mode').
Aug 13 16:41:45 crc-build-1 haproxy[57604]: [ALERT] 224/164145 (57604) : Fatal errors found in configuration.
Aug 13 16:41:45 crc-build-1 systemd[1]: haproxy.service: Control process exited, code=exited status=1
Aug 13 16:41:45 crc-build-1 systemd[1]: haproxy.service: Failed with result 'exit-code'.
Aug 13 16:41:45 crc-build-1 systemd[1]: Failed to start HAProxy Load Balancer.

...

$ haproxy version
HA-Proxy version 1.8.27-493ce0b 2020/11/06
Copyright 2000-2020 Willy Tarreau <[email protected]>

My OS:

cat /etc/os-release

NAME="CentOS Linux"
VERSION="8"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="8"
PLATFORM_ID="platform:el8"
PRETTY_NAME="CentOS Linux 8"
ANSI_COLOR="0;31"
CPE_NAME="cpe:/o:centos:centos:8"
HOME_URL="https://centos.org/"
BUG_REPORT_URL="https://bugs.centos.org/"
CENTOS_MANTISBT_PROJECT="CentOS-8"
CENTOS_MANTISBT_PROJECT_VERSION="8"