Refs. swh/infra/sysadm-environment#5389

Related to swh/infra/sysadm-environment#5389

Refs. swh/infra/sysadm-environment#5389

Related to swh/infra/sysadm-environment#5386

Related to swh/infra/sysadm-environment#5386

Related to swh/infra/sysadm-environment#5042

Related to swh/infra/sysadm-environment#5042

It will be active only after a restart of the nodes

Related to swh/infra/sysadm-environment#4824

Related to swh/infra/sysadm-environment#4919

This reverts commit 32116116.

We'll not use the rack affinity because our server dispatching
in the rack is not compatible with the cassandra needs.
It will also add some constraints on the cluster provisionning
we could not afford (need to keep the same number of servers in
all the racks)

Related to swh/infra/sysadm-environment#4801

Related to swh/infra/sysadm-environment#4801

Related to swh/infra/sysadm-environment#4800

Related to swh/infra/sysadm-environment#4711

Related to swh/infra/sysadm-environment#4711

It's necessary to disable all the options relation to the CMS
gc in the default file

Related to infra/sysadm-environment#4706

Related to infra/sysadm-environment#4691

It was 16G during the grid5000 experiment

Related to infra/sysadm-environment#4691

If remote access is activated, the user authentication is configured
and activated

Related to T4458

- Move the previous yaml base configuration to a puppet template
- Install cassandra via the archive instead of the debian packages
  It gives more flexibilities regarding the multi instances configuration
- Support multi instance per server
- Make the cassandra.yaml content less verbose

There is still some work to do:
- Manage more cassandra configuration properties
- Add the tcp port monitoring
- Wire the exporter metrics to one of the prometheus

Related to T4373