Cephadm on Ubuntu 22.04
Intro
Mission: Deploy new Ceph poduction-ready cluster on 4 new hardware nodes.
DevOps - OpenStack / K8S / CICD
Mission: Deploy new Ceph poduction-ready cluster on 4 new hardware nodes.
Mission: Create cpg / virtual volume / host / vlun via hp 3par cli and detect it in Linux
Mission: Debug failed github-pages site build (powered by JekyII).
Mission: Deploy DevStack release Queens on Virtual Machine (VirtualBox; Centos 7) and test custom roles/policies with Patrole project.
Mission: Deploy Openstack Kolla (release Train) with Ironic on 1 baremetal node with installed CentOS 7.
Why not Bifrost: We need all core openstack services (like nova/glance/neutron etc) and ability to use this baremetal node not only as Ironic controller but as normal AIO environment for infra instances (like nexus/gitlab/jenkins/etc vm’s).
Why Train: Right now (end 2019/beginning of 2020 ) we usually deploy release Stein. But Ironic Stein have some bugs like https://bugs.launchpad.net/kolla-ansible/+bug/1843067. So we decided to go with Train instead ;)
Mission: Install OpenStack clients in pythno2.7 venv in the year 2020
Why bother: Because of cmd2/more-itertools and other python3 deps you can’t just run “pip install python-openstackclient”.
Mission: Automate several actions in vmware cluster via python pyvmomi library.
Case: Delete/Create/Update several VM’s. Create/Delete/Revert to snapshots, etc…
Mission: Implement simple process usage monitoring via linux tools Case: Investigate huge rss(30+GB)/cpu usage in openstack-exporter
Mission: Deploy and operate VMware Harbor registry
Mission: Deploy Kolla-ansible release Train with 1 controller, 2 computes and external Ceph (business purpose - test bug with live-migration in Train);
Mission: Deploy ceph-ansible for further Openstack deployment. 3 hosts with collocated mons and osds.
Mission: Create custom centos/ubuntu cloud image with node-exporter and mtr-exporter for cloud SLA monitoring
Mission: Install ELK on VM , configure rsyslog export to Logstash -> ElasticSearch
Mission: Deploy standalone Prometheus instances for pulling metrics from several OpenShift clusters.
Mission: Deploy OpenShift OKD in isolated infrastructure.
Some golang stuff
Mission: create cull pronetheus openstack exporter
Several commands for NUMA administration
Always forget this stuff)
At some point, you will need building procedure for openstack packages (deb in our case). You can do it with following commands (I use Jenkins job for that).
We had one customer who wanted to deploy new Ceph cluster on two nodes, integrate them in Old Openstack Mirantis 7.0 environment and start migration process (instances, volumes, images, etc) to it.
At first glance this task seems pretty simple - just create new cinder backend, copy new ceph.conf/keyrings and begin migration on holidays/weekend.
But in reality, old openstack services such as glance,nova,cinder dont work with new rbd/rados in Ceph Luminous. But we found a way! So let’s start.
Simple benchmarking of gluster cluster with nfs-ganesha export and FIO tester.