Add persistent volumes
vigevenoj opened this issue · comments
This is currently fine running via minikube but to go live it needs to have persistent volumes for storing stuff, which so far is:
- influxdb data/config
- chronograf data/config
- kapacitor data/config
- telegraf data/config
- postgresql data
- traefik data
~30gb (5gb for each, even though it seems overkill for traefik) should be fine to start with, as long some sort of monitoring+alerting is set up for when that's no longer enough
This needs clean up for publishing but is mostly done:
- Volumes created and attached to VMs via the linode cli
- Gluster/Heketi spun up by running gk-deploy.sh from the gluster-kubernetes on one of the nodes in the cluster
The volume creation/attachment code is in the python that needs to be cleaned up and published
Ansible playbooks generate our topology.json, copy it to a node, and use it as input to gk-deploy.
All that's left to do is to either take the output of gk-deploy and scrape it for the heketi address, or use kubectl to do that, and then put that value into storageclass.yaml
Given that https://github.com/heketi/heketi is in "almost maintenance" mode and https://github.com/gluster/gluster-kubernetes is archived; don't use them for storage here. linode-blockstorage-csi looks like the best option going forward and lines up with other supported mechanisms