ceph/doc/cephadm/install.rst

============================
Deploying a new Ceph cluster
============================

Cephadm creates a new Ceph cluster by "bootstrapping" on a single
host, expanding the cluster to encompass any additional hosts, and
then deploying the needed services.

.. highlight:: console

Requirements
============

- Systemd
- Podman or Docker for running containers.
- Time synchronization (such as chrony or NTP)
- LVM2 for provisioning storage devices

Any modern Linux distribution should be sufficient.  Dependencies
are installed automatically by the bootstrap process below.


Get cephadm
===========

The ``cephadm`` command is used (1) to bootstrap a new cluster, (2) to
access a containerized shell with a working Ceph CLI, and (3) to work
with containerized Ceph daemons when debugging issues.

* You can use ``curl`` to fetch the most recent version of the
  standalone script::

    # curl --silent --remote-name --location https://github.com/ceph/ceph/raw/octopus/src/cephadm/cephadm
    # chmod +x cephadm

  This script can be run directly from the current directory with::

    # ./cephadm <arguments...>

* Although the standalone script is sufficient to get a cluster started, it is
  convenient to have the ``cephadm`` command installed on the host.  To install
  these packages for the current Octopus release::

    # ./cephadm add-repo --release octopus
    # ./cephadm install

  Confirm that ``cephadm`` is now in your PATH with::

    # which cephadm

* Some commercial Linux distributions (e.g., RHEL, SLE) may already
  include up-to-date Ceph packages.  In that case, you can install
  cephadm directly.  For example::

    # dnf install -y cephadm     # or
    # zypper install -y cephadm



Bootstrap a new cluster
=======================

You need to know which *IP address* to use for the cluster's first
monitor daemon.  This is normally just the IP for the first host.  If there
are multiple networks and interfaces, be sure to choose one that will
be accessible by any hosts accessing the Ceph cluster.

To bootstrap the cluster run the following commands::

  # mkdir -p /etc/ceph
  # cephadm bootstrap --mon-ip *<mon-ip>*

This command does a few things:

* A monitor and manager daemon for the new cluster are created on the
  local host.  A minimal configuration file needed to communicate with
  the new cluster is written to ``/etc/ceph/ceph.conf``.
* A copy of the ``client.admin`` administrative (privileged!) secret
  key is written to ``/etc/ceph/ceph.client.admin.keyring``.
* A new SSH key is generated for the Ceph cluster and is added to the
  root user's ``/root/.ssh/authorized_keys`` file.  A copy of the
  public key is written to ``/etc/ceph/ceph.pub``.

The default bootstrap behavior will work for the vast majority of
users.  See below for a few options that may be useful for some users,
or run ``cephadm bootstrap -h`` to see all available options:

* Bootstrap writes the files needed to access the new cluster to
  ``/etc/ceph`` for convenience, so that any Ceph packages installed
  on the host itself (e.g., to access the command line interface) can
  easily find them.

  Daemon containers deployed with cephadm, however, do not need
  ``/etc/ceph`` at all.  Use the ``--output-dir *<directory>*`` option
  to put them in a different directory (like ``.``), avoiding any
  potential conflicts with existing Ceph configuration (cephadm or
  otherwise) on the same host.

* You can pass any initial Ceph configuration options to the new
  cluster by putting them in a standard ini-style configuration file
  and using the ``--config *<config-file>*`` option.


Accessing the ceph command
==========================

Cephadm does not require any Ceph packages to be installed on the host.
However, it is often convenient to have something installed so that you
can access the ``ceph`` command.  There are several ways to do this:

* The ``cephadm shell`` command will launch a bash shell in a container.  By
  default, if configuration and keyring files are found in ``/etc/ceph`` on the
  host, they are passed into the container environment so that the shell
  is fully functional::

    # cephadm shell

* It may be helpful to create an alias::

    # alias ceph='cephadm shell --'

* You can install the ``ceph-common`` package, which contains all of the
  ceph commands, including ``ceph``, ``rbd``, ``mount.ceph`` (for mounting
  CephFS file systems), etc.::

    # cephadm install ceph-common

Confirm that the ``ceph`` command is accessible with::

  # ceph -v

Confirm that the ``ceph`` command can connect to the cluster and also
its status with::

  # ceph status


Adding hosts to the cluster
===========================

For each new host you'd like to add to the cluster, you need to do two things:

#. Install the cluster's public SSH key in the new host's root user's
   ``authorized_keys`` file::

     # ssh-copy-id -f -i ceph.pub root@*<new-host>*

   For example::

     # ssh-copy-id -f -i ceph.pub root@host2
     # ssh-copy-id -f -i ceph.pub root@host3

#. Tell Ceph that the new node is part of the cluster::

     # ceph orch host add *newhost*

   For example::

     # ceph orch host add host2
     # ceph orch host add host3


Deploying additional monitors (optional)
========================================

Normally a Ceph cluster has three or five monitor daemons spread
across different hosts.  As a rule of thumb, you should deploy five
monitors if there are five or more nodes in your cluster.

.. _CIDR: https://en.wikipedia.org/wiki/Classless_Inter-Domain_Routing#CIDR_notation

When Ceph knows what IP subnet the monitors should use it can automatically
deploy and scale monitors as the cluster grows (or contracts).  By default,
Ceph assumes that other monitors should use the same subnet that the first
monitor IP belongs to.

If your Ceph monitors (or the entire cluster) live on a single subnet,
then cephadm will automatically add up to give monitors as you add new
hosts to the cluster, and no further steps are necessary.

* If there is a specific IP subnet that should be used by monitors, you
  can configure that in `CIDR`_ format (e.g., ``10.1.2.0/24``) with::

    # ceph config set mon public_network *<mon-cidr-network>*

  For example::

    # ceph config set mon public_network 10.1.2.0/24

  Cephadm will only deploy new monitor daemons on hosts that have IPs
  configured in the configured subnet.

* If you want fewer (or more) than the default of 5 monitors::

    # ceph orch apply mon *<number-of-monitors>*

* To deploy monitors on a specific set of hosts::

    # ceph orch apply mon *<host1,host2,host3,...>*

  Be sure to include the first (bootstrap) host in this list.

* You can control which hosts the monitors run on by making use of
  host labels.  To set the ``mon`` label to the appropriate
  hosts::

    # ceph orch host label add *<hostname>* mon

  To view the current hosts and labels::

    # ceph orch host ls

  For example::

    # ceph orch host label add host1 mon
    # ceph orch host label add host2 mon
    # ceph orch host label add host3 mon
    # ceph orch host ls
    HOST   ADDR   LABELS  STATUS
    host1         mon
    host2         mon
    host3         mon
    host4
    host5

  Then tell cephadm to deploy monitors based on the label::

    # ceph orch apply mon label:mon

* You can explicitly specify the IP address or CIDR network for each monitor
  and control where it is placed.  To disable automated monitor deployment::

    # ceph orch apply mon --unmanaged

  To deploy each additional monitor::

    # ceph orch daemon add mon *<host1:ip-or-network1> [<host1:ip-or-network-2>...]*

  For example, to deploy a second monitor on ``newhost1`` using an IP
  address ``10.1.2.123`` and a third monitor on ``newhost2`` in
  network ``10.1.2.0/24``::

    # ceph orch apply mon --unmanaged
    # ceph orch daemon add mon newhost1:10.1.2.123
    # ceph orch daemon add mon newhost2:10.1.2.0/24


Deploying OSDs
==============

Any Ceph cluster needs OSDs (object storage daemons) to manage storage
devices (HDDs, SSDs) to store actual data for the cluster.

Storage devices on all cluster hosts can be displayed with::

  # ceph orch device ls

A storage device is considered *available* if all of the following
conditions are met.  Ceph is refuse to provision an OSD on a device
that is not available.

* The device must have no partitions.
* The device must not have any LVM state.
* The device must not be mounted.
* The device must not contain a file system.
* The device must not contain a Ceph BlueStore OSD.
* The device must be larger than 5 GB

There are a few ways to create new OSDs:

* The simplest way to configure storage devices is to tell Ceph to
  consume any available and unused storage device::

    # ceph orch apply osd --all-available-devices

* To create an OSD from a specific device on a specific host::

    # ceph orch daemon add osd *<host>*:*<device-path>*

  For example::

    # ceph orch daemon add osd host1:/dev/sdb

* You can use :ref:`drivegroups` to describe device(s) to consume
  based on their properties, such device type (SSD or HDD), device
  model names, size, or the hosts on which the devices exist::

    # ceph orch osd create -i spec.yml


Deploying MDSs
==============

One or more MDS daemons is required to use the CephFS file system.
These are created automatically if the newer ``ceph fs volume``
interface is used to create a new file system.  For more information,
see :ref:`fs-volumes-and-subvolumes`.

To deploy metadata servers::

  # ceph orch apply mds *<fs-name>* *<num-daemons>* [*<host1>* ...]

Deploying RGWs
==============

Cephadm deploys radosgw as a collection of daemons that manage a
particular *realm* and *zone*.  (For more information about realms and
zones, see :ref:`multisite`.)  To deploy a set of radosgw daemons for
a particular realm and zone::

  # ceph orch apply rgw *<realm-name>* *<zone-name>* *<num-daemons>* [*<host1>* ...]

Note that with cephadm, radosgw daemons are configured via the monitor
configuration database instead of via a `ceph.conf` or the command line.  If
that confiruation isn't already in place (usually in the
``client.rgw.<realmname>.<zonename>`` section), then the radosgw
daemons will start up with default settings (e.g., binding to port
80).