diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-27 18:24:20 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-27 18:24:20 +0000 |
commit | 483eb2f56657e8e7f419ab1a4fab8dce9ade8609 (patch) | |
tree | e5d88d25d870d5dedacb6bbdbe2a966086a0a5cf /doc/ceph-volume | |
parent | Initial commit. (diff) | |
download | ceph-483eb2f56657e8e7f419ab1a4fab8dce9ade8609.tar.xz ceph-483eb2f56657e8e7f419ab1a4fab8dce9ade8609.zip |
Adding upstream version 14.2.21.upstream/14.2.21upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
-rw-r--r-- | doc/ceph-volume/index.rst | 83 | ||||
-rw-r--r-- | doc/ceph-volume/intro.rst | 84 | ||||
-rw-r--r-- | doc/ceph-volume/inventory.rst | 17 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/activate.rst | 113 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/batch.rst | 129 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/create.rst | 24 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/encryption.rst | 86 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/index.rst | 28 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/list.rst | 184 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/prepare.rst | 332 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/scan.rst | 9 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/systemd.rst | 28 | ||||
-rw-r--r-- | doc/ceph-volume/lvm/zap.rst | 65 | ||||
-rw-r--r-- | doc/ceph-volume/simple/activate.rst | 80 | ||||
-rw-r--r-- | doc/ceph-volume/simple/index.rst | 32 | ||||
-rw-r--r-- | doc/ceph-volume/simple/scan.rst | 176 | ||||
-rw-r--r-- | doc/ceph-volume/simple/systemd.rst | 28 | ||||
-rw-r--r-- | doc/ceph-volume/systemd.rst | 49 | ||||
-rw-r--r-- | doc/ceph-volume/zfs/index.rst | 31 | ||||
-rw-r--r-- | doc/ceph-volume/zfs/inventory.rst | 19 |
20 files changed, 1597 insertions, 0 deletions
diff --git a/doc/ceph-volume/index.rst b/doc/ceph-volume/index.rst new file mode 100644 index 00000000..36ce9cc8 --- /dev/null +++ b/doc/ceph-volume/index.rst @@ -0,0 +1,83 @@ +.. _ceph-volume: + +ceph-volume +=========== +Deploy OSDs with different device technologies like lvm or physical disks using +pluggable tools (:doc:`lvm/index` itself is treated like a plugin) and trying to +follow a predictable, and robust way of preparing, activating, and starting OSDs. + +:ref:`Overview <ceph-volume-overview>` | +:ref:`Plugin Guide <ceph-volume-plugins>` | + + +**Command Line Subcommands** + +There is currently support for ``lvm``, and plain disks (with GPT partitions) +that may have been deployed with ``ceph-disk``. + +``zfs`` support is available for running a FreeBSD cluster. + +* :ref:`ceph-volume-lvm` +* :ref:`ceph-volume-simple` +* :ref:`ceph-volume-zfs` + +**Node inventory** + +The :ref:`ceph-volume-inventory` subcommand provides information and metadata +about a nodes physical disk inventory. + + +Migrating +--------- +Starting on Ceph version 13.0.0, ``ceph-disk`` is deprecated. Deprecation +warnings will show up that will link to this page. It is strongly suggested +that users start consuming ``ceph-volume``. There are two paths for migrating: + +#. Keep OSDs deployed with ``ceph-disk``: The :ref:`ceph-volume-simple` command + provides a way to take over the management while disabling ``ceph-disk`` + triggers. +#. Redeploy existing OSDs with ``ceph-volume``: This is covered in depth on + :ref:`rados-replacing-an-osd` + +For details on why ``ceph-disk`` was removed please see the :ref:`Why was +ceph-disk replaced? <ceph-disk-replaced>` section. + + +New deployments +^^^^^^^^^^^^^^^ +For new deployments, :ref:`ceph-volume-lvm` is recommended, it can use any +logical volume as input for data OSDs, or it can setup a minimal/naive logical +volume from a device. + +Existing OSDs +^^^^^^^^^^^^^ +If the cluster has OSDs that were provisioned with ``ceph-disk``, then +``ceph-volume`` can take over the management of these with +:ref:`ceph-volume-simple`. A scan is done on the data device or OSD directory, +and ``ceph-disk`` is fully disabled. Encryption is fully supported. + + +.. toctree:: + :hidden: + :maxdepth: 3 + :caption: Contents: + + intro + systemd + inventory + lvm/index + lvm/activate + lvm/batch + lvm/encryption + lvm/prepare + lvm/create + lvm/scan + lvm/systemd + lvm/list + lvm/zap + simple/index + simple/activate + simple/scan + simple/systemd + zfs/index + zfs/inventory diff --git a/doc/ceph-volume/intro.rst b/doc/ceph-volume/intro.rst new file mode 100644 index 00000000..d59dd2b6 --- /dev/null +++ b/doc/ceph-volume/intro.rst @@ -0,0 +1,84 @@ +.. _ceph-volume-overview: + +Overview +-------- +The ``ceph-volume`` tool aims to be a single purpose command line tool to deploy +logical volumes as OSDs, trying to maintain a similar API to ``ceph-disk`` when +preparing, activating, and creating OSDs. + +It deviates from ``ceph-disk`` by not interacting or relying on the udev rules +that come installed for Ceph. These rules allow automatic detection of +previously setup devices that are in turn fed into ``ceph-disk`` to activate +them. + +.. _ceph-disk-replaced: + +Replacing ``ceph-disk`` +----------------------- +The ``ceph-disk`` tool was created at a time were the project was required to +support many different types of init systems (upstart, sysvinit, etc...) while +being able to discover devices. This caused the tool to concentrate initially +(and exclusively afterwards) on GPT partitions. Specifically on GPT GUIDs, +which were used to label devices in a unique way to answer questions like: + +* is this device a Journal? +* an encrypted data partition? +* was the device left partially prepared? + +To solve these, it used ``UDEV`` rules to match the GUIDs, that would call +``ceph-disk``, and end up in a back and forth between the ``ceph-disk`` systemd +unit and the ``ceph-disk`` executable. The process was very unreliable and time +consuming (a timeout of close to three hours **per OSD** had to be put in +place), and would cause OSDs to not come up at all during the boot process of +a node. + +It was hard to debug, or even replicate these problems given the asynchronous +behavior of ``UDEV``. + +Since the world-view of ``ceph-disk`` had to be GPT partitions exclusively, it meant +that it couldn't work with other technologies like LVM, or similar device +mapper devices. It was ultimately decided to create something modular, starting +with LVM support, and the ability to expand on other technologies as needed. + + +GPT partitions are simple? +-------------------------- +Although partitions in general are simple to reason about, ``ceph-disk`` +partitions were not simple by any means. It required a tremendous amount of +special flags in order to get them to work correctly with the device discovery +workflow. Here is an example call to create a data partition:: + + /sbin/sgdisk --largest-new=1 --change-name=1:ceph data --partition-guid=1:f0fc39fd-eeb2-49f1-b922-a11939cf8a0f --typecode=1:89c57f98-2fe5-4dc0-89c1-f3ad0ceff2be --mbrtogpt -- /dev/sdb + +Not only creating these was hard, but these partitions required devices to be +exclusively owned by Ceph. For example, in some cases a special partition would +be created when devices were encrypted, which would contain unencrypted keys. +This was ``ceph-disk`` domain knowledge, which would not translate to a "GPT +partitions are simple" understanding. Here is an example of that special +partition being created:: + + /sbin/sgdisk --new=5:0:+10M --change-name=5:ceph lockbox --partition-guid=5:None --typecode=5:fb3aabf9-d25f-47cc-bf5e-721d181642be --mbrtogpt -- /dev/sdad + + +Modularity +---------- +``ceph-volume`` was designed to be a modular tool because we anticipate that +there are going to be lots of ways that people provision the hardware devices +that we need to consider. There are already two: legacy ceph-disk devices that +are still in use and have GPT partitions (handled by :ref:`ceph-volume-simple`), +and lvm. SPDK devices where we manage NVMe devices directly from userspace are +on the immediate horizon, where LVM won't work there since the kernel isn't +involved at all. + +``ceph-volume lvm`` +------------------- +By making use of :term:`LVM tags`, the :ref:`ceph-volume-lvm` sub-command is +able to store and later re-discover and query devices associated with OSDs so +that they can later be activated. + +LVM performance penalty +----------------------- +In short: we haven't been able to notice any significant performance penalties +associated with the change to LVM. By being able to work closely with LVM, the +ability to work with other device mapper technologies was a given: there is no +technical difficulty in working with anything that can sit below a Logical Volume. diff --git a/doc/ceph-volume/inventory.rst b/doc/ceph-volume/inventory.rst new file mode 100644 index 00000000..edb1fd20 --- /dev/null +++ b/doc/ceph-volume/inventory.rst @@ -0,0 +1,17 @@ +.. _ceph-volume-inventory: + +``inventory`` +============= +The ``inventory`` subcommand queries a host's disc inventory and provides +hardware information and metadata on every physical device. + +By default the command returns a short, human-readable report of all physical disks. + +For programmatic consumption of this report pass ``--format json`` to generate a +JSON formatted report. This report includes extensive information on the +physical drives such as disk metadata (like model and size), logical volumes +and whether they are used by ceph, and if the disk is usable by ceph and +reasons why not. + +A device path can be specified to report extensive information on a device in +both plain and json format. diff --git a/doc/ceph-volume/lvm/activate.rst b/doc/ceph-volume/lvm/activate.rst new file mode 100644 index 00000000..5e43d4cb --- /dev/null +++ b/doc/ceph-volume/lvm/activate.rst @@ -0,0 +1,113 @@ +.. _ceph-volume-lvm-activate: + +``activate`` +============ +Once :ref:`ceph-volume-lvm-prepare` is completed, and all the various steps +that entails are done, the volume is ready to get "activated". + +This activation process enables a systemd unit that persists the OSD ID and its +UUID (also called ``fsid`` in Ceph CLI tools), so that at boot time it can +understand what OSD is enabled and needs to be mounted. + +.. note:: The execution of this call is fully idempotent, and there is no + side-effects when running multiple times + +New OSDs +-------- +To activate newly prepared OSDs both the :term:`OSD id` and :term:`OSD uuid` +need to be supplied. For example:: + + ceph-volume lvm activate --bluestore 0 0263644D-0BF1-4D6D-BC34-28BD98AE3BC8 + +.. note:: The UUID is stored in the ``fsid`` file in the OSD path, which is + generated when :ref:`ceph-volume-lvm-prepare` is used. + +Activating all OSDs +------------------- +It is possible to activate all existing OSDs at once by using the ``--all`` +flag. For example:: + + ceph-volume lvm activate --all + +This call will inspect all the OSDs created by ceph-volume that are inactive +and will activate them one by one. If any of the OSDs are already running, it +will report them in the command output and skip them, making it safe to rerun +(idempotent). + +requiring uuids +^^^^^^^^^^^^^^^ +The :term:`OSD uuid` is being required as an extra step to ensure that the +right OSD is being activated. It is entirely possible that a previous OSD with +the same id exists and would end up activating the incorrect one. + + +dmcrypt +^^^^^^^ +If the OSD was prepared with dmcrypt by ceph-volume, there is no need to +specify ``--dmcrypt`` on the command line again (that flag is not available for +the ``activate`` subcommand). An encrypted OSD will be automatically detected. + + +Discovery +--------- +With OSDs previously created by ``ceph-volume``, a *discovery* process is +performed using :term:`LVM tags` to enable the systemd units. + +The systemd unit will capture the :term:`OSD id` and :term:`OSD uuid` and +persist it. Internally, the activation will enable it like:: + + systemctl enable ceph-volume@lvm-$id-$uuid + +For example:: + + systemctl enable ceph-volume@lvm-0-8715BEB4-15C5-49DE-BA6F-401086EC7B41 + +Would start the discovery process for the OSD with an id of ``0`` and a UUID of +``8715BEB4-15C5-49DE-BA6F-401086EC7B41``. + +.. note:: for more details on the systemd workflow see :ref:`ceph-volume-lvm-systemd` + +The systemd unit will look for the matching OSD device, and by looking at its +:term:`LVM tags` will proceed to: + +# mount the device in the corresponding location (by convention this is + ``/var/lib/ceph/osd/<cluster name>-<osd id>/``) + +# ensure that all required devices are ready for that OSD. In the case of +a journal (when ``--filestore`` is selected) the device will be queried (with +``blkid`` for partitions, and lvm for logical volumes) to ensure that the +correct device is being linked. The symbolic link will *always* be re-done to +ensure that the correct device is linked. + +# start the ``ceph-osd@0`` systemd unit + +.. note:: The system infers the objectstore type (filestore or bluestore) by + inspecting the LVM tags applied to the OSD devices + +Existing OSDs +------------- +For existing OSDs that have been deployed with ``ceph-disk``, they need to be +scanned and activated :ref:`using the simple sub-command <ceph-volume-simple>`. +If a different tooling was used then the only way to port them over to the new +mechanism is to prepare them again (losing data). See +:ref:`ceph-volume-lvm-existing-osds` for details on how to proceed. + +Summary +------- +To recap the ``activate`` process for :term:`bluestore`: + +#. require both :term:`OSD id` and :term:`OSD uuid` +#. enable the system unit with matching id and uuid +#. Create the ``tmpfs`` mount at the OSD directory in + ``/var/lib/ceph/osd/$cluster-$id/`` +#. Recreate all the files needed with ``ceph-bluestore-tool prime-osd-dir`` by + pointing it to the OSD ``block`` device. +#. the systemd unit will ensure all devices are ready and linked +#. the matching ``ceph-osd`` systemd unit will get started + +And for :term:`filestore`: + +#. require both :term:`OSD id` and :term:`OSD uuid` +#. enable the system unit with matching id and uuid +#. the systemd unit will ensure all devices are ready and mounted (if needed) +#. the matching ``ceph-osd`` systemd unit will get started diff --git a/doc/ceph-volume/lvm/batch.rst b/doc/ceph-volume/lvm/batch.rst new file mode 100644 index 00000000..84959fb5 --- /dev/null +++ b/doc/ceph-volume/lvm/batch.rst @@ -0,0 +1,129 @@ +.. _ceph-volume-lvm-batch: + +``batch`` +=========== +The subcommand allows to create multiple OSDs at the same time given +an input of devices. + +The subcommand is based to :ref:`ceph-volume-lvm-create`, and will use the very +same code path. All ``batch`` does is to calculate the appropriate sizes of all +volumes and skip over already created volumes. + +All the features that ``ceph-volume lvm create`` supports, like ``dmcrypt``, +avoiding ``systemd`` units from starting, defining bluestore or filestore, +are supported. + + +.. _ceph-volume-lvm-batch_auto: + +Automatic sorting of disks +-------------------------- +If ``batch`` receives only a single list of devices and the ``--no-auto`` option + is not passed, ``ceph-volume`` will auto-sort disks by its rotational + property and use non-rotating disks for ``block.db`` or ``journal`` depending + on the objectstore used. +This behavior is now DEPRECATED and will be removed in future releases. Instead + an ``auto`` option will be introduced to retain this behavior. +It is recommended to make use of the explicit device lists for ``block.db``, + ``block.wal`` and ``journal``. +For example assuming :term:`bluestore` is used and ``--no-auto`` is not passed, + the deprecated behavior would deploy the following, depending on the devices + passed: + +#. Devices are all spinning HDDs: 1 OSD is created per device +#. Devices are all SSDs: 2 OSDs are created per device +#. Devices are a mix of HDDs and SSDs: data is placed on the spinning device, + the ``block.db`` is created on the SSD, as large as possible. + +.. note:: Although operations in ``ceph-volume lvm create`` allow usage of + ``block.wal`` it isn't supported with the ``auto`` behavior. + +.. _ceph-volume-lvm-batch_bluestore: + +Reporting +========= +By default ``batch`` will print a report of the computed OSD layout and ask the +user to confirm. This can be overridden by passing ``--yes``. + +If one wants to try out several invocations with being asked to deploy +``--report`` can be passed. ``ceph-volume`` will exit after printing the report. + +Consider the following invocation:: + + $ ceph-volume lvm batch --report /dev/sdb /dev/sdc /dev/sdd --db-devices /dev/nvme0n1 + +This will deploy three OSDs with external ``db`` and ``wal`` volumes on +an NVME device. + +**pretty reporting** +The ``pretty`` report format (the default) would +look like this:: + + $ ceph-volume lvm batch --report /dev/sdb /dev/sdc /dev/sdd --db-devices /dev/nvme0n1 + + Total OSDs: 3 + + + + +**JSON reporting** +Reporting can produce a structured output with ``--format json``:: + + $ ceph-volume lvm batch --report /dev/sdb /dev/sdc /dev/sdd --db-devices /dev/nvme0n1 + +Sizing +====== +When no sizing arguments are passed, `ceph-volume` will derive the sizing from +the passed device lists (or the sorted lists when using the automatic sorting). +`ceph-volume batch` will attempt to fully utilize a devices available capacity. +Relying on automatic sizing is recommended. + +If one requires a different sizing policy for wal, db or journal devices, +`ceph-volume` offers implicit and explicit sizing rules. + +Implicit sizing +--------------- +Scenarios in which either devices are under-comitted or not all data devices are +currently ready for use (due to a broken disk for example), one can still rely +on `ceph-volume` automatic sizing. +Users can provide hints to `ceph-volume` as to how many data devices should have +their external volumes on a set of fast devices. These options are: + +* `--block-db-slots` +* `--block-wal-slots` +* `--journal-slots` + +For example consider an OSD host that is supposed to contain 5 data devices and +one device for wal/db volumes. However one data device is currently broken and +is being replaced. Instead of calculating the explicit sizes for the wal/db +volume one can simply call:: + + $ ceph-volume lvm batch --report /dev/sdb /dev/sdc /dev/sdd /dev/sde --db-devices /dev/nvme0n1 --block-db-slots 5 + +Explicit sizing +--------------- +It is also possible to provide explicit sizes to `ceph-volume` via the arguments + +* `--block-db-size` +* `--block-wal-size` +* `--journal-size` + +`ceph-volume` will try to satisfy the requested sizes given the passed disks. If +this is not possible, no OSDs will be deployed. + + +Idempotency and disk replacements +================================= +`ceph-volume lvm batch` intends to be idempotent, i.e. calling the same command +repeatedly must result in the same outcome. For example calling:: + + $ ceph-volume lvm batch --report /dev/sdb /dev/sdc /dev/sdd --db-devices /dev/nvme0n1 + +will result in three deployed OSDs (if all disks were available). Calling this +command again, you will still end up with three OSDs and ceph-volume will exit +with return code 0. + +Suppose /dev/sdc goes bad and needs to be replaced. After destroying the OSD and +replacing the hardware, you can again call the same command and `ceph-volume` +will detect that only two out of the three wanted OSDs are setup and re-create +the missing OSD. diff --git a/doc/ceph-volume/lvm/create.rst b/doc/ceph-volume/lvm/create.rst new file mode 100644 index 00000000..c90d1f6f --- /dev/null +++ b/doc/ceph-volume/lvm/create.rst @@ -0,0 +1,24 @@ +.. _ceph-volume-lvm-create: + +``create`` +=========== +This subcommand wraps the two-step process to provision a new osd (calling +``prepare`` first and then ``activate``) into a single +one. The reason to prefer ``prepare`` and then ``activate`` is to gradually +introduce new OSDs into a cluster, and avoiding large amounts of data being +rebalanced. + +The single-call process unifies exactly what :ref:`ceph-volume-lvm-prepare` and +:ref:`ceph-volume-lvm-activate` do, with the convenience of doing it all at +once. + +There is nothing different to the process except the OSD will become up and in +immediately after completion. + +The backing objectstore can be specified with: + +* :ref:`--filestore <ceph-volume-lvm-prepare_filestore>` +* :ref:`--bluestore <ceph-volume-lvm-prepare_bluestore>` + +All command line flags and options are the same as ``ceph-volume lvm prepare``. +Please refer to :ref:`ceph-volume-lvm-prepare` for details. diff --git a/doc/ceph-volume/lvm/encryption.rst b/doc/ceph-volume/lvm/encryption.rst new file mode 100644 index 00000000..1483ef32 --- /dev/null +++ b/doc/ceph-volume/lvm/encryption.rst @@ -0,0 +1,86 @@ +.. _ceph-volume-lvm-encryption: + +Encryption +========== + +Logical volumes can be encrypted using ``dmcrypt`` by specifying the +``--dmcrypt`` flag when creating OSDs. Encryption can be done in different ways, +specially with LVM. ``ceph-volume`` is somewhat opinionated with the way it +sets up encryption with logical volumes so that the process is consistent and +robust. + +In this case, ``ceph-volume lvm`` follows these constraints: + +* only LUKS (version 1) is used +* Logical Volumes are encrypted, while their underlying PVs (physical volumes) + aren't +* Non-LVM devices like partitions are also encrypted with the same OSD key + + +LUKS +---- +There are currently two versions of LUKS, 1 and 2. Version 2 is a bit easier +to implement but not widely available in all distros Ceph supports. LUKS 1 is +not going to be deprecated in favor of LUKS 2, so in order to have as wide +support as possible, ``ceph-volume`` uses LUKS version 1. + +.. note:: Version 1 of LUKS is just referenced as "LUKS" whereas version 2 is + referred to as LUKS2 + + +LUKS on LVM +----------- +Encryption is done on top of existing logical volumes (unlike encrypting the +physical device). Any single logical volume can be encrypted while other +volumes can remain unencrypted. This method also allows for flexible logical +volume setups, since encryption will happen once the LV is created. + + +Workflow +-------- +When setting up the OSD, a secret key will be created, that will be passed +along to the monitor in JSON format as ``stdin`` to prevent the key from being +captured in the logs. + +The JSON payload looks something like:: + + { + "cephx_secret": CEPHX_SECRET, + "dmcrypt_key": DMCRYPT_KEY, + "cephx_lockbox_secret": LOCKBOX_SECRET, + } + +The naming convention for the keys is **strict**, and they are named like that +for the hardcoded (legacy) names ceph-disk used. + +* ``cephx_secret`` : The cephx key used to authenticate +* ``dmcrypt_key`` : The secret (or private) key to unlock encrypted devices +* ``cephx_lockbox_secret`` : The authentication key used to retrieve the + ``dmcrypt_key``. It is named *lockbox* because ceph-disk used to have an + unencrypted partition named after it, used to store public keys and other + OSD metadata. + +The naming convention is strict because Monitors supported the naming +convention by ceph-disk, which used these key names. In order to keep +compatibility and prevent ceph-disk from breaking, ceph-volume will use the same +naming convention *although they don't make sense for the new encryption +workflow*. + +After the common steps of setting up the OSD during the prepare stage, either +with :term:`filestore` or :term:`bluestore`, the logical volume is left ready +to be activated, regardless of the state of the device (encrypted or decrypted). + +At activation time, the logical volume will get decrypted and the OSD started +once the process completes correctly. + +Summary of the encryption workflow for creating a new OSD: + +#. OSD is created, both lockbox and dmcrypt keys are created, and sent along + with JSON to the monitors, indicating an encrypted OSD. + +#. All complementary devices (like journal, db, or wal) get created and + encrypted with the same OSD key. Key is stored in the LVM metadata of the + OSD + +#. Activation continues by ensuring devices are mounted, retrieving the dmcrypt + secret key from the monitors and decrypting before the OSD gets started. diff --git a/doc/ceph-volume/lvm/index.rst b/doc/ceph-volume/lvm/index.rst new file mode 100644 index 00000000..9a2191fb --- /dev/null +++ b/doc/ceph-volume/lvm/index.rst @@ -0,0 +1,28 @@ +.. _ceph-volume-lvm: + +``lvm`` +======= +Implements the functionality needed to deploy OSDs from the ``lvm`` subcommand: +``ceph-volume lvm`` + +**Command Line Subcommands** + +* :ref:`ceph-volume-lvm-prepare` + +* :ref:`ceph-volume-lvm-activate` + +* :ref:`ceph-volume-lvm-create` + +* :ref:`ceph-volume-lvm-list` + +.. not yet implemented +.. * :ref:`ceph-volume-lvm-scan` + +**Internal functionality** + +There are other aspects of the ``lvm`` subcommand that are internal and not +exposed to the user, these sections explain how these pieces work together, +clarifying the workflows of the tool. + +:ref:`Systemd Units <ceph-volume-lvm-systemd>` | +:ref:`lvm <ceph-volume-lvm-api>` diff --git a/doc/ceph-volume/lvm/list.rst b/doc/ceph-volume/lvm/list.rst new file mode 100644 index 00000000..718154b1 --- /dev/null +++ b/doc/ceph-volume/lvm/list.rst @@ -0,0 +1,184 @@ +.. _ceph-volume-lvm-list: + +``list`` +======== +This subcommand will list any devices (logical and physical) that may be +associated with a Ceph cluster, as long as they contain enough metadata to +allow for that discovery. + +Output is grouped by the OSD ID associated with the devices, and unlike +``ceph-disk`` it does not provide any information for devices that aren't +associated with Ceph. + +Command line options: + +* ``--format`` Allows a ``json`` or ``pretty`` value. Defaults to ``pretty`` + which will group the device information in a human-readable format. + +Full Reporting +-------------- +When no positional arguments are used, a full reporting will be presented. This +means that all devices and logical volumes found in the system will be +displayed. + +Full ``pretty`` reporting for two OSDs, one with a lv as a journal, and another +one with a physical device may look similar to:: + + # ceph-volume lvm list + + + ====== osd.1 ======= + + [journal] /dev/journals/journal1 + + journal uuid C65n7d-B1gy-cqX3-vZKY-ZoE0-IEYM-HnIJzs + osd id 1 + cluster fsid ce454d91-d748-4751-a318-ff7f7aa18ffd + type journal + osd fsid 661b24f8-e062-482b-8110-826ffe7f13fa + data uuid SlEgHe-jX1H-QBQk-Sce0-RUls-8KlY-g8HgcZ + journal device /dev/journals/journal1 + data device /dev/test_group/data-lv2 + devices /dev/sda + + [data] /dev/test_group/data-lv2 + + journal uuid C65n7d-B1gy-cqX3-vZKY-ZoE0-IEYM-HnIJzs + osd id 1 + cluster fsid ce454d91-d748-4751-a318-ff7f7aa18ffd + type data + osd fsid 661b24f8-e062-482b-8110-826ffe7f13fa + data uuid SlEgHe-jX1H-QBQk-Sce0-RUls-8KlY-g8HgcZ + journal device /dev/journals/journal1 + data device /dev/test_group/data-lv2 + devices /dev/sdb + + ====== osd.0 ======= + + [data] /dev/test_group/data-lv1 + + journal uuid cd72bd28-002a-48da-bdf6-d5b993e84f3f + osd id 0 + cluster fsid ce454d91-d748-4751-a318-ff7f7aa18ffd + type data + osd fsid 943949f0-ce37-47ca-a33c-3413d46ee9ec + data uuid TUpfel-Q5ZT-eFph-bdGW-SiNW-l0ag-f5kh00 + journal device /dev/sdd1 + data device /dev/test_group/data-lv1 + devices /dev/sdc + + [journal] /dev/sdd1 + + PARTUUID cd72bd28-002a-48da-bdf6-d5b993e84f3f + + +For logical volumes the ``devices`` key is populated with the physical devices +associated with the logical volume. Since LVM allows multiple physical devices +to be part of a logical volume, the value will be comma separated when using +``pretty``, but an array when using ``json``. + +.. note:: Tags are displayed in a readable format. The ``osd id`` key is stored + as a ``ceph.osd_id`` tag. For more information on lvm tag conventions + see :ref:`ceph-volume-lvm-tag-api` + +Single Reporting +---------------- +Single reporting can consume both devices and logical volumes as input +(positional parameters). For logical volumes, it is required to use the group +name as well as the logical volume name. + +For example the ``data-lv2`` logical volume, in the ``test_group`` volume group +can be listed in the following way:: + + # ceph-volume lvm list test_group/data-lv2 + + + ====== osd.1 ======= + + [data] /dev/test_group/data-lv2 + + journal uuid C65n7d-B1gy-cqX3-vZKY-ZoE0-IEYM-HnIJzs + osd id 1 + cluster fsid ce454d91-d748-4751-a318-ff7f7aa18ffd + type data + osd fsid 661b24f8-e062-482b-8110-826ffe7f13fa + data uuid SlEgHe-jX1H-QBQk-Sce0-RUls-8KlY-g8HgcZ + journal device /dev/journals/journal1 + data device /dev/test_group/data-lv2 + devices /dev/sdc + + +.. note:: Tags are displayed in a readable format. The ``osd id`` key is stored + as a ``ceph.osd_id`` tag. For more information on lvm tag conventions + see :ref:`ceph-volume-lvm-tag-api` + + +For plain disks, the full path to the device is required. For example, for +a device like ``/dev/sdd1`` it can look like:: + + + # ceph-volume lvm list /dev/sdd1 + + + ====== osd.0 ======= + + [journal] /dev/sdd1 + + PARTUUID cd72bd28-002a-48da-bdf6-d5b993e84f3f + + + +``json`` output +--------------- +All output using ``--format=json`` will show everything the system has stored +as metadata for the devices, including tags. + +No changes for readability are done with ``json`` reporting, and all +information is presented as-is. Full output as well as single devices can be +listed. + +For brevity, this is how a single logical volume would look with ``json`` +output (note how tags aren't modified):: + + # ceph-volume lvm list --format=json test_group/data-lv1 + { + "0": [ + { + "devices": ["/dev/sda"], + "lv_name": "data-lv1", + "lv_path": "/dev/test_group/data-lv1", + "lv_tags": "ceph.cluster_fsid=ce454d91-d748-4751-a318-ff7f7aa18ffd,ceph.data_device=/dev/test_group/data-lv1,ceph.data_uuid=TUpfel-Q5ZT-eFph-bdGW-SiNW-l0ag-f5kh00,ceph.journal_device=/dev/sdd1,ceph.journal_uuid=cd72bd28-002a-48da-bdf6-d5b993e84f3f,ceph.osd_fsid=943949f0-ce37-47ca-a33c-3413d46ee9ec,ceph.osd_id=0,ceph.type=data", + "lv_uuid": "TUpfel-Q5ZT-eFph-bdGW-SiNW-l0ag-f5kh00", + "name": "data-lv1", + "path": "/dev/test_group/data-lv1", + "tags": { + "ceph.cluster_fsid": "ce454d91-d748-4751-a318-ff7f7aa18ffd", + "ceph.data_device": "/dev/test_group/data-lv1", + "ceph.data_uuid": "TUpfel-Q5ZT-eFph-bdGW-SiNW-l0ag-f5kh00", + "ceph.journal_device": "/dev/sdd1", + "ceph.journal_uuid": "cd72bd28-002a-48da-bdf6-d5b993e84f3f", + "ceph.osd_fsid": "943949f0-ce37-47ca-a33c-3413d46ee9ec", + "ceph.osd_id": "0", + "ceph.type": "data" + }, + "type": "data", + "vg_name": "test_group" + } + ] + } + + +Synchronized information +------------------------ +Before any listing type, the lvm API is queried to ensure that physical devices +that may be in use haven't changed naming. It is possible that non-persistent +devices like ``/dev/sda1`` could change to ``/dev/sdb1``. + +The detection is possible because the ``PARTUUID`` is stored as part of the +metadata in the logical volume for the data lv. Even in the case of a journal +that is a physical device, this information is still stored on the data logical +volume associated with it. + +If the name is no longer the same (as reported by ``blkid`` when using the +``PARTUUID``), the tag will get updated and the report will use the newly +refreshed information. diff --git a/doc/ceph-volume/lvm/prepare.rst b/doc/ceph-volume/lvm/prepare.rst new file mode 100644 index 00000000..78a7c091 --- /dev/null +++ b/doc/ceph-volume/lvm/prepare.rst @@ -0,0 +1,332 @@ +.. _ceph-volume-lvm-prepare: + +``prepare`` +=========== +This subcommand allows a :term:`filestore` or :term:`bluestore` setup. It is +recommended to pre-provision a logical volume before using it with +``ceph-volume lvm``. + +Logical volumes are not altered except for adding extra metadata. + +.. note:: This is part of a two step process to deploy an OSD. If looking for + a single-call way, please see :ref:`ceph-volume-lvm-create` + +To help identify volumes, the process of preparing a volume (or volumes) to +work with Ceph, the tool will assign a few pieces of metadata information using +:term:`LVM tags`. + +:term:`LVM tags` makes volumes easy to discover later, and help identify them as +part of a Ceph system, and what role they have (journal, filestore, bluestore, +etc...) + +Although initially :term:`filestore` is supported (and supported by default) +the back end can be specified with: + + +* :ref:`--filestore <ceph-volume-lvm-prepare_filestore>` +* :ref:`--bluestore <ceph-volume-lvm-prepare_bluestore>` + +.. _ceph-volume-lvm-prepare_bluestore: + +``bluestore`` +------------- +The :term:`bluestore` objectstore is the default for new OSDs. It offers a bit +more flexibility for devices compared to :term:`filestore`. +Bluestore supports the following configurations: + +* A block device, a block.wal, and a block.db device +* A block device and a block.wal device +* A block device and a block.db device +* A single block device + +The bluestore subcommand accepts physical block devices, partitions on +physical block devices or logical volumes as arguments for the various device parameters +If a physical device is provided, a logical volume will be created. A volume group will +either be created or reused it its name begins with ``ceph``. +This allows a simpler approach at using LVM but at the cost of flexibility: +there are no options or configurations to change how the LV is created. + +The ``block`` is specified with the ``--data`` flag, and in its simplest use +case it looks like:: + + ceph-volume lvm prepare --bluestore --data vg/lv + +A raw device can be specified in the same way:: + + ceph-volume lvm prepare --bluestore --data /path/to/device + +For enabling :ref:`encryption <ceph-volume-lvm-encryption>`, the ``--dmcrypt`` flag is required:: + + ceph-volume lvm prepare --bluestore --dmcrypt --data vg/lv + +If a ``block.db`` or a ``block.wal`` is needed (they are optional for +bluestore) they can be specified with ``--block.db`` and ``--block.wal`` +accordingly. These can be a physical device, a partition or +a logical volume. + +For both ``block.db`` and ``block.wal`` partitions aren't made logical volumes +because they can be used as-is. + +While creating the OSD directory, the process will use a ``tmpfs`` mount to +place all the files needed for the OSD. These files are initially created by +``ceph-osd --mkfs`` and are fully ephemeral. + +A symlink is always created for the ``block`` device, and optionally for +``block.db`` and ``block.wal``. For a cluster with a default name, and an OSD +id of 0, the directory could look like:: + + # ls -l /var/lib/ceph/osd/ceph-0 + lrwxrwxrwx. 1 ceph ceph 93 Oct 20 13:05 block -> /dev/ceph-be2b6fbd-bcf2-4c51-b35d-a35a162a02f0/osd-block-25cf0a05-2bc6-44ef-9137-79d65bd7ad62 + lrwxrwxrwx. 1 ceph ceph 93 Oct 20 13:05 block.db -> /dev/sda1 + lrwxrwxrwx. 1 ceph ceph 93 Oct 20 13:05 block.wal -> /dev/ceph/osd-wal-0 + -rw-------. 1 ceph ceph 37 Oct 20 13:05 ceph_fsid + -rw-------. 1 ceph ceph 37 Oct 20 13:05 fsid + -rw-------. 1 ceph ceph 55 Oct 20 13:05 keyring + -rw-------. 1 ceph ceph 6 Oct 20 13:05 ready + -rw-------. 1 ceph ceph 10 Oct 20 13:05 type + -rw-------. 1 ceph ceph 2 Oct 20 13:05 whoami + +In the above case, a device was used for ``block`` so ``ceph-volume`` create +a volume group and a logical volume using the following convention: + +* volume group name: ``ceph-{cluster fsid}`` or if the vg exists already + ``ceph-{random uuid}`` + +* logical volume name: ``osd-block-{osd_fsid}`` + + +.. _ceph-volume-lvm-prepare_filestore: + +``filestore`` +------------- +This is the OSD backend that allows preparation of logical volumes for +a :term:`filestore` objectstore OSD. + +It can use a logical volume for the OSD data and a physical device, a partition +or logical volume for the journal. A physical device will have a logical volume +created on it. A volume group will either be created or reused it its name begins +with ``ceph``. No special preparation is needed for these volumes other than +following the minimum size requirements for data and journal. + +The CLI call looks like this of a basic standalone filestore OSD:: + + ceph-volume lvm prepare --filestore --data <data block device> + +To deploy file store with an external journal:: + + ceph-volume lvm prepare --filestore --data <data block device> --journal <journal block device> + +For enabling :ref:`encryption <ceph-volume-lvm-encryption>`, the ``--dmcrypt`` flag is required:: + + ceph-volume lvm prepare --filestore --dmcrypt --data <data block device> --journal <journal block device> + +Both the journal and data block device can take three forms: + +* a physical block device +* a partition on a physical block device +* a logical volume + +When using logical volumes the value *must* be of the format +``volume_group/logical_volume``. Since logical volume names +are not enforced for uniqueness, this prevents accidentally +choosing the wrong volume. + +When using a partition, it *must* contain a ``PARTUUID``, that can be +discovered by ``blkid``. THis ensure it can later be identified correctly +regardless of the device name (or path). + +For example: passing a logical volume for data and a partition ``/dev/sdc1`` for +the journal:: + + ceph-volume lvm prepare --filestore --data volume_group/lv_name --journal /dev/sdc1 + +Passing a bare device for data and a logical volume ias the journal:: + + ceph-volume lvm prepare --filestore --data /dev/sdc --journal volume_group/journal_lv + +A generated uuid is used to ask the cluster for a new OSD. These two pieces are +crucial for identifying an OSD and will later be used throughout the +:ref:`ceph-volume-lvm-activate` process. + +The OSD data directory is created using the following convention:: + + /var/lib/ceph/osd/<cluster name>-<osd id> + +At this point the data volume is mounted at this location, and the journal +volume is linked:: + + ln -s /path/to/journal /var/lib/ceph/osd/<cluster_name>-<osd-id>/journal + +The monmap is fetched using the bootstrap key from the OSD:: + + /usr/bin/ceph --cluster ceph --name client.bootstrap-osd + --keyring /var/lib/ceph/bootstrap-osd/ceph.keyring + mon getmap -o /var/lib/ceph/osd/<cluster name>-<osd id>/activate.monmap + +``ceph-osd`` will be called to populate the OSD directory, that is already +mounted, re-using all the pieces of information from the initial steps:: + + ceph-osd --cluster ceph --mkfs --mkkey -i <osd id> \ + --monmap /var/lib/ceph/osd/<cluster name>-<osd id>/activate.monmap --osd-data \ + /var/lib/ceph/osd/<cluster name>-<osd id> --osd-journal /var/lib/ceph/osd/<cluster name>-<osd id>/journal \ + --osd-uuid <osd uuid> --keyring /var/lib/ceph/osd/<cluster name>-<osd id>/keyring \ + --setuser ceph --setgroup ceph + + +.. _ceph-volume-lvm-partitions: + +Partitioning +------------ +``ceph-volume lvm`` does not currently create partitions from a whole device. +If using device partitions the only requirement is that they contain the +``PARTUUID`` and that it is discoverable by ``blkid``. Both ``fdisk`` and +``parted`` will create that automatically for a new partition. + +For example, using a new, unformatted drive (``/dev/sdd`` in this case) we can +use ``parted`` to create a new partition. First we list the device +information:: + + $ parted --script /dev/sdd print + Model: VBOX HARDDISK (scsi) + Disk /dev/sdd: 11.5GB + Sector size (logical/physical): 512B/512B + Disk Flags: + +This device is not even labeled yet, so we can use ``parted`` to create +a ``gpt`` label before we create a partition, and verify again with ``parted +print``:: + + $ parted --script /dev/sdd mklabel gpt + $ parted --script /dev/sdd print + Model: VBOX HARDDISK (scsi) + Disk /dev/sdd: 11.5GB + Sector size (logical/physical): 512B/512B + Partition Table: gpt + Disk Flags: + +Now lets create a single partition, and verify later if ``blkid`` can find +a ``PARTUUID`` that is needed by ``ceph-volume``:: + + $ parted --script /dev/sdd mkpart primary 1 100% + $ blkid /dev/sdd1 + /dev/sdd1: PARTLABEL="primary" PARTUUID="16399d72-1e1f-467d-96ee-6fe371a7d0d4" + + +.. _ceph-volume-lvm-existing-osds: + +Existing OSDs +------------- +For existing clusters that want to use this new system and have OSDs that are +already running there are a few things to take into account: + +.. warning:: this process will forcefully format the data device, destroying + existing data, if any. + +* OSD paths should follow this convention:: + + /var/lib/ceph/osd/<cluster name>-<osd id> + +* Preferably, no other mechanisms to mount the volume should exist, and should + be removed (like fstab mount points) + +The one time process for an existing OSD, with an ID of 0 and using +a ``"ceph"`` cluster name would look like (the following command will **destroy +any data** in the OSD):: + + ceph-volume lvm prepare --filestore --osd-id 0 --osd-fsid E3D291C1-E7BF-4984-9794-B60D9FA139CB + +The command line tool will not contact the monitor to generate an OSD ID and +will format the LVM device in addition to storing the metadata on it so that it +can be started later (for detailed metadata description see +:ref:`ceph-volume-lvm-tags`). + + +Crush device class +------------------ + +To set the crush device class for the OSD, use the ``--crush-device-class`` flag. This will +work for both bluestore and filestore OSDs:: + + ceph-volume lvm prepare --bluestore --data vg/lv --crush-device-class foo + + +.. _ceph-volume-lvm-multipath: + +``multipath`` support +--------------------- +Devices that come from ``multipath`` are not supported as-is. The tool will +refuse to consume a raw multipath device and will report a message like:: + + --> RuntimeError: Cannot use device (/dev/mapper/<name>). A vg/lv path or an existing device is needed + +The reason for not supporting multipath is that depending on the type of the +multipath setup, if using an active/passive array as the underlying physical +devices, filters are required in ``lvm.conf`` to exclude the disks that are part of +those underlying devices. + +It is unfeasible for ceph-volume to understand what type of configuration is +needed for LVM to be able to work in various different multipath scenarios. The +functionality to create the LV for you is merely a (naive) convenience, +anything that involves different settings or configuration must be provided by +a config management system which can then provide VGs and LVs for ceph-volume +to consume. + +This situation will only arise when trying to use the ceph-volume functionality +that creates a volume group and logical volume from a device. If a multipath +device is already a logical volume it *should* work, given that the LVM +configuration is done correctly to avoid issues. + + +Storing metadata +---------------- +The following tags will get applied as part of the preparation process +regardless of the type of volume (journal or data) or OSD objectstore: + +* ``cluster_fsid`` +* ``encrypted`` +* ``osd_fsid`` +* ``osd_id`` +* ``crush_device_class`` + +For :term:`filestore` these tags will be added: + +* ``journal_device`` +* ``journal_uuid`` + +For :term:`bluestore` these tags will be added: + +* ``block_device`` +* ``block_uuid`` +* ``db_device`` +* ``db_uuid`` +* ``wal_device`` +* ``wal_uuid`` + +.. note:: For the complete lvm tag conventions see :ref:`ceph-volume-lvm-tag-api` + + +Summary +------- +To recap the ``prepare`` process for :term:`bluestore`: + +#. Accepts raw physical devices, partitions on physical devices or logical volumes as arguments. +#. Creates logical volumes on any raw physical devices. +#. Generate a UUID for the OSD +#. Ask the monitor get an OSD ID reusing the generated UUID +#. OSD data directory is created on a tmpfs mount. +#. ``block``, ``block.wal``, and ``block.db`` are symlinked if defined. +#. monmap is fetched for activation +#. Data directory is populated by ``ceph-osd`` +#. Logical Volumes are assigned all the Ceph metadata using lvm tags + + +And the ``prepare`` process for :term:`filestore`: + +#. Accepts raw physical devices, partitions on physical devices or logical volumes as arguments. +#. Generate a UUID for the OSD +#. Ask the monitor get an OSD ID reusing the generated UUID +#. OSD data directory is created and data volume mounted +#. Journal is symlinked from data volume to journal location +#. monmap is fetched for activation +#. devices is mounted and data directory is populated by ``ceph-osd`` +#. data and journal volumes are assigned all the Ceph metadata using lvm tags diff --git a/doc/ceph-volume/lvm/scan.rst b/doc/ceph-volume/lvm/scan.rst new file mode 100644 index 00000000..aa9990f7 --- /dev/null +++ b/doc/ceph-volume/lvm/scan.rst @@ -0,0 +1,9 @@ +scan +==== +This sub-command will allow to discover Ceph volumes previously setup by the +tool by looking into the system's logical volumes and their tags. + +As part of the :ref:`ceph-volume-lvm-prepare` process, the logical volumes are assigned +a few tags with important pieces of information. + +.. note:: This sub-command is not yet implemented diff --git a/doc/ceph-volume/lvm/systemd.rst b/doc/ceph-volume/lvm/systemd.rst new file mode 100644 index 00000000..30260de7 --- /dev/null +++ b/doc/ceph-volume/lvm/systemd.rst @@ -0,0 +1,28 @@ +.. _ceph-volume-lvm-systemd: + +systemd +======= +Upon startup, it will identify the logical volume using :term:`LVM tags`, +finding a matching ID and later ensuring it is the right one with +the :term:`OSD uuid`. + +After identifying the correct volume it will then proceed to mount it by using +the OSD destination conventions, that is:: + + /var/lib/ceph/osd/<cluster name>-<osd id> + +For our example OSD with an id of ``0``, that means the identified device will +be mounted at:: + + + /var/lib/ceph/osd/ceph-0 + + +Once that process is complete, a call will be made to start the OSD:: + + systemctl start ceph-osd@0 + +The systemd portion of this process is handled by the ``ceph-volume lvm +trigger`` sub-command, which is only in charge of parsing metadata coming from +systemd and startup, and then dispatching to ``ceph-volume lvm activate`` which +would proceed with activation. diff --git a/doc/ceph-volume/lvm/zap.rst b/doc/ceph-volume/lvm/zap.rst new file mode 100644 index 00000000..367d7469 --- /dev/null +++ b/doc/ceph-volume/lvm/zap.rst @@ -0,0 +1,65 @@ +.. _ceph-volume-lvm-zap: + +``zap`` +======= + +This subcommand is used to zap lvs, partitions or raw devices that have been used +by ceph OSDs so that they may be reused. If given a path to a logical +volume it must be in the format of vg/lv. Any filesystems present +on the given lv or partition will be removed and all data will be purged. + +.. note:: The lv or partition will be kept intact. + +.. note:: If the logical volume, raw device or partition is being used for any ceph related + mount points they will be unmounted. + +Zapping a logical volume:: + + ceph-volume lvm zap {vg name/lv name} + +Zapping a partition:: + + ceph-volume lvm zap /dev/sdc1 + +Removing Devices +---------------- +When zapping, and looking for full removal of the device (lv, vg, or partition) +use the ``--destroy`` flag. A common use case is to simply deploy OSDs using +a whole raw device. If you do so and then wish to reuse that device for another +OSD you must use the ``--destroy`` flag when zapping so that the vgs and lvs +that ceph-volume created on the raw device will be removed. + +.. note:: Multiple devices can be accepted at once, to zap them all + +Zapping a raw device and destroying any vgs or lvs present:: + + ceph-volume lvm zap /dev/sdc --destroy + + +This action can be performed on partitions, and logical volumes as well:: + + ceph-volume lvm zap /dev/sdc1 --destroy + ceph-volume lvm zap osd-vg/data-lv --destroy + + +Finally, multiple devices can be detected if filtering by OSD ID and/or OSD +FSID. Either identifier can be used or both can be used at the same time. This +is useful in situations where multiple devices associated with a specific ID +need to be purged. When using the FSID, the filtering is stricter, and might +not match other (possibly invalid) devices associated to an ID. + +By ID only:: + + ceph-volume lvm zap --destroy --osd-id 1 + +By FSID:: + + ceph-volume lvm zap --destroy --osd-fsid 2E8FBE58-0328-4E3B-BFB7-3CACE4E9A6CE + +By both:: + + ceph-volume lvm zap --destroy --osd-fsid 2E8FBE58-0328-4E3B-BFB7-3CACE4E9A6CE --osd-id 1 + + +.. warning:: If the systemd unit associated with the OSD ID to be zapped is + detected as running, the tool will refuse to zap until the daemon is stopped. diff --git a/doc/ceph-volume/simple/activate.rst b/doc/ceph-volume/simple/activate.rst new file mode 100644 index 00000000..edbb1e3f --- /dev/null +++ b/doc/ceph-volume/simple/activate.rst @@ -0,0 +1,80 @@ +.. _ceph-volume-simple-activate: + +``activate`` +============ +Once :ref:`ceph-volume-simple-scan` has been completed, and all the metadata +captured for an OSD has been persisted to ``/etc/ceph/osd/{id}-{uuid}.json`` +the OSD is now ready to get "activated". + +This activation process **disables** all ``ceph-disk`` systemd units by masking +them, to prevent the UDEV/ceph-disk interaction that will attempt to start them +up at boot time. + +The disabling of ``ceph-disk`` units is done only when calling ``ceph-volume +simple activate`` directly, but is is avoided when being called by systemd when +the system is booting up. + +The activation process requires using both the :term:`OSD id` and :term:`OSD uuid` +To activate parsed OSDs:: + + ceph-volume simple activate 0 6cc43680-4f6e-4feb-92ff-9c7ba204120e + +The above command will assume that a JSON configuration will be found in:: + + /etc/ceph/osd/0-6cc43680-4f6e-4feb-92ff-9c7ba204120e.json + +Alternatively, using a path to a JSON file directly is also possible:: + + ceph-volume simple activate --file /etc/ceph/osd/0-6cc43680-4f6e-4feb-92ff-9c7ba204120e.json + +requiring uuids +^^^^^^^^^^^^^^^ +The :term:`OSD uuid` is being required as an extra step to ensure that the +right OSD is being activated. It is entirely possible that a previous OSD with +the same id exists and would end up activating the incorrect one. + + +Discovery +--------- +With OSDs previously scanned by ``ceph-volume``, a *discovery* process is +performed using ``blkid`` and ``lvm``. There is currently support only for +devices with GPT partitions and LVM logical volumes. + +The GPT partitions will have a ``PARTUUID`` that can be queried by calling out +to ``blkid``, and the logical volumes will have a ``lv_uuid`` that can be +queried against ``lvs`` (the LVM tool to list logical volumes). + +This discovery process ensures that devices can be correctly detected even if +they are repurposed into another system or if their name changes (as in the +case of non-persisting names like ``/dev/sda1``) + +The JSON configuration file used to map what devices go to what OSD will then +coordinate the mounting and symlinking as part of activation. + +To ensure that the symlinks are always correct, if they exist in the OSD +directory, the symlinks will be re-done. + +A systemd unit will capture the :term:`OSD id` and :term:`OSD uuid` and +persist it. Internally, the activation will enable it like:: + + systemctl enable ceph-volume@simple-$id-$uuid + +For example:: + + systemctl enable ceph-volume@simple-0-8715BEB4-15C5-49DE-BA6F-401086EC7B41 + +Would start the discovery process for the OSD with an id of ``0`` and a UUID of +``8715BEB4-15C5-49DE-BA6F-401086EC7B41``. + + +The systemd process will call out to activate passing the information needed to +identify the OSD and its devices, and it will proceed to: + +# mount the device in the corresponding location (by convention this is + ``/var/lib/ceph/osd/<cluster name>-<osd id>/``) + +# ensure that all required devices are ready for that OSD and properly linked, +regardless of objectstore used (filestore or bluestore). The symbolic link will +**always** be re-done to ensure that the correct device is linked. + +# start the ``ceph-osd@0`` systemd unit diff --git a/doc/ceph-volume/simple/index.rst b/doc/ceph-volume/simple/index.rst new file mode 100644 index 00000000..315dea99 --- /dev/null +++ b/doc/ceph-volume/simple/index.rst @@ -0,0 +1,32 @@ +.. _ceph-volume-simple: + +``simple`` +========== +Implements the functionality needed to manage OSDs from the ``simple`` subcommand: +``ceph-volume simple`` + +**Command Line Subcommands** + +* :ref:`ceph-volume-simple-scan` + +* :ref:`ceph-volume-simple-activate` + +* :ref:`ceph-volume-simple-systemd` + + +By *taking over* management, it disables all ``ceph-disk`` systemd units used +to trigger devices at startup, relying on basic (customizable) JSON +configuration and systemd for starting up OSDs. + +This process involves two steps: + +#. :ref:`Scan <ceph-volume-simple-scan>` the running OSD or the data device +#. :ref:`Activate <ceph-volume-simple-activate>` the scanned OSD + +The scanning will infer everything that ``ceph-volume`` needs to start the OSD, +so that when activation is needed, the OSD can start normally without getting +interference from ``ceph-disk``. + +As part of the activation process the systemd units for ``ceph-disk`` in charge +of reacting to ``udev`` events, are linked to ``/dev/null`` so that they are +fully inactive. diff --git a/doc/ceph-volume/simple/scan.rst b/doc/ceph-volume/simple/scan.rst new file mode 100644 index 00000000..2749b14b --- /dev/null +++ b/doc/ceph-volume/simple/scan.rst @@ -0,0 +1,176 @@ +.. _ceph-volume-simple-scan: + +``scan`` +======== +Scanning allows to capture any important details from an already-deployed OSD +so that ``ceph-volume`` can manage it without the need of any other startup +workflows or tools (like ``udev`` or ``ceph-disk``). Encryption with LUKS or +PLAIN formats is fully supported. + +The command has the ability to inspect a running OSD, by inspecting the +directory where the OSD data is stored, or by consuming the data partition. +The command can also scan all running OSDs if no path or device is provided. + +Once scanned, information will (by default) persist the metadata as JSON in +a file in ``/etc/ceph/osd``. This ``JSON`` file will use the naming convention +of: ``{OSD ID}-{OSD FSID}.json``. An OSD with an id of 1, and an FSID like +``86ebd829-1405-43d3-8fd6-4cbc9b6ecf96`` the absolute path of the file would +be:: + + /etc/ceph/osd/1-86ebd829-1405-43d3-8fd6-4cbc9b6ecf96.json + +The ``scan`` subcommand will refuse to write to this file if it already exists. +If overwriting the contents is needed, the ``--force`` flag must be used:: + + ceph-volume simple scan --force {path} + +If there is no need to persist the ``JSON`` metadata, there is support to send +the contents to ``stdout`` (no file will be written):: + + ceph-volume simple scan --stdout {path} + + +.. _ceph-volume-simple-scan-directory: + +Running OSDs scan +----------------- +Using this command without providing an OSD directory or device will scan the +directories of any currently running OSDs. If a running OSD was not created +by ceph-disk it will be ignored and not scanned. + +To scan all running ceph-disk OSDs, the command would look like:: + + ceph-volume simple scan + +Directory scan +-------------- +The directory scan will capture OSD file contents from interesting files. There +are a few files that must exist in order to have a successful scan: + +* ``ceph_fsid`` +* ``fsid`` +* ``keyring`` +* ``ready`` +* ``type`` +* ``whoami`` + +If the OSD is encrypted, it will additionally add the following keys: + +* ``encrypted`` +* ``encryption_type`` +* ``lockbox_keyring`` + +In the case of any other file, as long as it is not a binary or a directory, it +will also get captured and persisted as part of the JSON object. + +The convention for the keys in the JSON object is that any file name will be +a key, and its contents will be its value. If the contents are a single line +(like in the case of the ``whoami``) the contents are trimmed, and the newline +is dropped. For example with an OSD with an id of 1, this is how the JSON entry +would look like:: + + "whoami": "1", + +For files that may have more than one line, the contents are left as-is, except +for keyrings which are treated specially and parsed to extract the keyring. For +example, a ``keyring`` that gets read as:: + + [osd.1]\n\tkey = AQBBJ/dZp57NIBAAtnuQS9WOS0hnLVe0rZnE6Q==\n + +Would get stored as:: + + "keyring": "AQBBJ/dZp57NIBAAtnuQS9WOS0hnLVe0rZnE6Q==", + + +For a directory like ``/var/lib/ceph/osd/ceph-1``, the command could look +like:: + + ceph-volume simple scan /var/lib/ceph/osd/ceph1 + + +.. _ceph-volume-simple-scan-device: + +Device scan +----------- +When an OSD directory is not available (OSD is not running, or device is not +mounted) the ``scan`` command is able to introspect the device to capture +required data. Just like :ref:`ceph-volume-simple-scan-directory`, it would +still require a few files present. This means that the device to be scanned +**must be** the data partition of the OSD. + +As long as the data partition of the OSD is being passed in as an argument, the +sub-command can scan its contents. + +In the case where the device is already mounted, the tool can detect this +scenario and capture file contents from that directory. + +If the device is not mounted, a temporary directory will be created, and the +device will be mounted temporarily just for scanning the contents. Once +contents are scanned, the device will be unmounted. + +For a device like ``/dev/sda1`` which **must** be a data partition, the command +could look like:: + + ceph-volume simple scan /dev/sda1 + + +.. _ceph-volume-simple-scan-json: + +``JSON`` contents +----------------- +The contents of the JSON object is very simple. The scan not only will persist +information from the special OSD files and their contents, but will also +validate paths and device UUIDs. Unlike what ``ceph-disk`` would do, by storing +them in ``{device type}_uuid`` files, the tool will persist them as part of the +device type key. + +For example, a ``block.db`` device would look something like:: + + "block.db": { + "path": "/dev/disk/by-partuuid/6cc43680-4f6e-4feb-92ff-9c7ba204120e", + "uuid": "6cc43680-4f6e-4feb-92ff-9c7ba204120e" + }, + +But it will also persist the ``ceph-disk`` special file generated, like so:: + + "block.db_uuid": "6cc43680-4f6e-4feb-92ff-9c7ba204120e", + +This duplication is in place because the tool is trying to ensure the +following: + +# Support OSDs that may not have ceph-disk special files +# Check the most up-to-date information on the device, by querying against LVM +and ``blkid`` +# Support both logical volumes and GPT devices + +This is a sample ``JSON`` metadata, from an OSD that is using ``bluestore``:: + + { + "active": "ok", + "block": { + "path": "/dev/disk/by-partuuid/40fd0a64-caa5-43a3-9717-1836ac661a12", + "uuid": "40fd0a64-caa5-43a3-9717-1836ac661a12" + }, + "block.db": { + "path": "/dev/disk/by-partuuid/6cc43680-4f6e-4feb-92ff-9c7ba204120e", + "uuid": "6cc43680-4f6e-4feb-92ff-9c7ba204120e" + }, + "block.db_uuid": "6cc43680-4f6e-4feb-92ff-9c7ba204120e", + "block_uuid": "40fd0a64-caa5-43a3-9717-1836ac661a12", + "bluefs": "1", + "ceph_fsid": "c92fc9eb-0610-4363-aafc-81ddf70aaf1b", + "cluster_name": "ceph", + "data": { + "path": "/dev/sdr1", + "uuid": "86ebd829-1405-43d3-8fd6-4cbc9b6ecf96" + }, + "fsid": "86ebd829-1405-43d3-8fd6-4cbc9b6ecf96", + "keyring": "AQBBJ/dZp57NIBAAtnuQS9WOS0hnLVe0rZnE6Q==", + "kv_backend": "rocksdb", + "magic": "ceph osd volume v026", + "mkfs_done": "yes", + "ready": "ready", + "systemd": "", + "type": "bluestore", + "whoami": "3" + } diff --git a/doc/ceph-volume/simple/systemd.rst b/doc/ceph-volume/simple/systemd.rst new file mode 100644 index 00000000..aa5bebff --- /dev/null +++ b/doc/ceph-volume/simple/systemd.rst @@ -0,0 +1,28 @@ +.. _ceph-volume-simple-systemd: + +systemd +======= +Upon startup, it will identify the logical volume by loading the JSON file in +``/etc/ceph/osd/{id}-{uuid}.json`` corresponding to the instance name of the +systemd unit. + +After identifying the correct volume it will then proceed to mount it by using +the OSD destination conventions, that is:: + + /var/lib/ceph/osd/{cluster name}-{osd id} + +For our example OSD with an id of ``0``, that means the identified device will +be mounted at:: + + + /var/lib/ceph/osd/ceph-0 + + +Once that process is complete, a call will be made to start the OSD:: + + systemctl start ceph-osd@0 + +The systemd portion of this process is handled by the ``ceph-volume simple +trigger`` sub-command, which is only in charge of parsing metadata coming from +systemd and startup, and then dispatching to ``ceph-volume simple activate`` which +would proceed with activation. diff --git a/doc/ceph-volume/systemd.rst b/doc/ceph-volume/systemd.rst new file mode 100644 index 00000000..6cbc1121 --- /dev/null +++ b/doc/ceph-volume/systemd.rst @@ -0,0 +1,49 @@ +.. _ceph-volume-systemd: + +systemd +======= +As part of the activation process (either with :ref:`ceph-volume-lvm-activate` +or :ref:`ceph-volume-simple-activate`), systemd units will get enabled that +will use the OSD id and uuid as part of their name. These units will be run +when the system boots, and will proceed to activate their corresponding +volumes via their sub-command implementation. + +The API for activation is a bit loose, it only requires two parts: the +subcommand to use and any extra meta information separated by a dash. This +convention makes the units look like:: + + ceph-volume@{command}-{extra metadata} + +The *extra metadata* can be anything needed that the subcommand implementing +the processing might need. In the case of :ref:`ceph-volume-lvm` and +:ref:`ceph-volume-simple`, both look to consume the :term:`OSD id` and :term:`OSD uuid`, +but this is not a hard requirement, it is just how the sub-commands are +implemented. + +Both the command and extra metadata gets persisted by systemd as part of the +*"instance name"* of the unit. For example an OSD with an ID of 0, for the +``lvm`` sub-command would look like:: + + systemctl enable ceph-volume@lvm-0-0A3E1ED2-DA8A-4F0E-AA95-61DEC71768D6 + +The enabled unit is a :term:`systemd oneshot` service, meant to start at boot +after the local filesystem is ready to be used. + + +Failure and Retries +------------------- +It is common to have failures when a system is coming up online. The devices +are sometimes not fully available and this unpredictable behavior may cause an +OSD to not be ready to be used. + +There are two configurable environment variables used to set the retry +behavior: + +* ``CEPH_VOLUME_SYSTEMD_TRIES``: Defaults to 30 +* ``CEPH_VOLUME_SYSTEMD_INTERVAL``: Defaults to 5 + +The *"tries"* is a number that sets the maximum amount of times the unit will +attempt to activate an OSD before giving up. + +The *"interval"* is a value in seconds that determines the waiting time before +initiating another try at activating the OSD. diff --git a/doc/ceph-volume/zfs/index.rst b/doc/ceph-volume/zfs/index.rst new file mode 100644 index 00000000..c06228de --- /dev/null +++ b/doc/ceph-volume/zfs/index.rst @@ -0,0 +1,31 @@ +.. _ceph-volume-zfs: + +``zfs`` +======= +Implements the functionality needed to deploy OSDs from the ``zfs`` subcommand: +``ceph-volume zfs`` + +The current implementation only works for ZFS on FreeBSD + +**Command Line Subcommands** + +* :ref:`ceph-volume-zfs-inventory` + +.. not yet implemented +.. * :ref:`ceph-volume-zfs-prepare` + +.. * :ref:`ceph-volume-zfs-activate` + +.. * :ref:`ceph-volume-zfs-create` + +.. * :ref:`ceph-volume-zfs-list` + +.. * :ref:`ceph-volume-zfs-scan` + +**Internal functionality** + +There are other aspects of the ``zfs`` subcommand that are internal and not +exposed to the user, these sections explain how these pieces work together, +clarifying the workflows of the tool. + +:ref:`zfs <ceph-volume-zfs-api>` diff --git a/doc/ceph-volume/zfs/inventory.rst b/doc/ceph-volume/zfs/inventory.rst new file mode 100644 index 00000000..fd00325b --- /dev/null +++ b/doc/ceph-volume/zfs/inventory.rst @@ -0,0 +1,19 @@ +.. _ceph-volume-zfs-inventory: + +``inventory`` +============= +The ``inventory`` subcommand queries a host's disc inventory through GEOM and provides +hardware information and metadata on every physical device. + +This only works on a FreeBSD platform. + +By default the command returns a short, human-readable report of all physical disks. + +For programmatic consumption of this report pass ``--format json`` to generate a +JSON formatted report. This report includes extensive information on the +physical drives such as disk metadata (like model and size), logical volumes +and whether they are used by ceph, and if the disk is usable by ceph and +reasons why not. + +A device path can be specified to report extensive information on a device in +both plain and json format. |