10 pveceph - Manage Ceph Services on Proxmox VE Nodes
15 include::pveceph.1-synopsis.adoc[]
21 Manage Ceph Services on Proxmox VE Nodes
22 ========================================
26 [thumbnail="screenshot/gui-ceph-status.png"]
28 {pve} unifies your compute and storage systems, i.e. you can use the same
29 physical nodes within a cluster for both computing (processing VMs and
30 containers) and replicated storage. The traditional silos of compute and
31 storage resources can be wrapped up into a single hyper-converged appliance.
32 Separate storage networks (SANs) and connections via network attached storages
33 (NAS) disappear. With the integration of Ceph, an open source software-defined
34 storage platform, {pve} has the ability to run and manage Ceph storage directly
35 on the hypervisor nodes.
37 Ceph is a distributed object store and file system designed to provide
38 excellent performance, reliability and scalability.
40 .Some advantages of Ceph on {pve} are:
41 - Easy setup and management with CLI and GUI support
45 - Scalable to the exabyte level
46 - Setup pools with different performance and redundancy characteristics
47 - Data is replicated, making it fault tolerant
48 - Runs on economical commodity hardware
49 - No need for hardware RAID controllers
52 For small to mid sized deployments, it is possible to install a Ceph server for
53 RADOS Block Devices (RBD) directly on your {pve} cluster nodes, see
54 xref:ceph_rados_block_devices[Ceph RADOS Block Devices (RBD)]. Recent
55 hardware has plenty of CPU power and RAM, so running storage services
56 and VMs on the same node is possible.
58 To simplify management, we provide 'pveceph' - a tool to install and
59 manage {ceph} services on {pve} nodes.
61 .Ceph consists of a couple of Daemons footnote:[Ceph intro http://docs.ceph.com/docs/luminous/start/intro/], for use as a RBD storage:
62 - Ceph Monitor (ceph-mon)
63 - Ceph Manager (ceph-mgr)
64 - Ceph OSD (ceph-osd; Object Storage Daemon)
66 TIP: We recommend to get familiar with the Ceph vocabulary.
67 footnote:[Ceph glossary http://docs.ceph.com/docs/luminous/glossary]
73 To build a Proxmox Ceph Cluster there should be at least three (preferably)
74 identical servers for the setup.
76 A 10Gb network, exclusively used for Ceph, is recommended. A meshed network
77 setup is also an option if there are no 10Gb switches available, see our wiki
78 article footnote:[Full Mesh Network for Ceph {webwiki-url}Full_Mesh_Network_for_Ceph_Server] .
80 Check also the recommendations from
81 http://docs.ceph.com/docs/luminous/start/hardware-recommendations/[Ceph's website].
84 As Ceph handles data object redundancy and multiple parallel writes to disks
85 (OSDs) on its own, using a RAID controller normally doesn’t improve
86 performance or availability. On the contrary, Ceph is designed to handle whole
87 disks on it's own, without any abstraction in between. RAID controller are not
88 designed for the Ceph use case and may complicate things and sometimes even
89 reduce performance, as their write and caching algorithms may interfere with
92 WARNING: Avoid RAID controller, use host bus adapter (HBA) instead.
96 Installation of Ceph Packages
97 -----------------------------
99 On each node run the installation script as follows:
106 This sets up an `apt` package repository in
107 `/etc/apt/sources.list.d/ceph.list` and installs the required software.
110 Creating initial Ceph configuration
111 -----------------------------------
113 [thumbnail="screenshot/gui-ceph-config.png"]
115 After installation of packages, you need to create an initial Ceph
116 configuration on just one node, based on your network (`10.10.10.0/24`
117 in the following example) dedicated for Ceph:
121 pveceph init --network 10.10.10.0/24
124 This creates an initial configuration at `/etc/pve/ceph.conf`. That file is
125 automatically distributed to all {pve} nodes by using
126 xref:chapter_pmxcfs[pmxcfs]. The command also creates a symbolic link
127 from `/etc/ceph/ceph.conf` pointing to that file. So you can simply run
128 Ceph commands without the need to specify a configuration file.
131 [[pve_ceph_monitors]]
132 Creating Ceph Monitors
133 ----------------------
135 [thumbnail="screenshot/gui-ceph-monitor.png"]
137 The Ceph Monitor (MON)
138 footnote:[Ceph Monitor http://docs.ceph.com/docs/luminous/start/intro/]
139 maintains a master copy of the cluster map. For high availability you need to
140 have at least 3 monitors.
142 On each node where you want to place a monitor (three monitors are recommended),
143 create it by using the 'Ceph -> Monitor' tab in the GUI or run.
151 This will also install the needed Ceph Manager ('ceph-mgr') by default. If you
152 do not want to install a manager, specify the '-exclude-manager' option.
156 Creating Ceph Manager
157 ----------------------
159 The Manager daemon runs alongside the monitors, providing an interface for
160 monitoring the cluster. Since the Ceph luminous release the
161 ceph-mgr footnote:[Ceph Manager http://docs.ceph.com/docs/luminous/mgr/] daemon
162 is required. During monitor installation the ceph manager will be installed as
165 NOTE: It is recommended to install the Ceph Manager on the monitor nodes. For
166 high availability install more then one manager.
178 [thumbnail="screenshot/gui-ceph-osd-status.png"]
180 via GUI or via CLI as follows:
184 pveceph createosd /dev/sd[X]
187 TIP: We recommend a Ceph cluster size, starting with 12 OSDs, distributed evenly
188 among your, at least three nodes (4 OSDs on each node).
190 If the disk was used before (eg. ZFS/RAID/OSD), to remove partition table, boot
191 sector and any OSD leftover the following commands should be sufficient.
195 dd if=/dev/zero of=/dev/sd[X] bs=1M count=200
196 ceph-disk zap /dev/sd[X]
199 WARNING: The above commands will destroy data on the disk!
204 Starting with the Ceph Kraken release, a new Ceph OSD storage type was
205 introduced, the so called Bluestore
206 footnote:[Ceph Bluestore http://ceph.com/community/new-luminous-bluestore/].
207 This is the default when creating OSDs in Ceph luminous.
211 pveceph createosd /dev/sd[X]
214 NOTE: In order to select a disk in the GUI, to be more fail-safe, the disk needs
215 to have a GPT footnoteref:[GPT, GPT partition table
216 https://en.wikipedia.org/wiki/GUID_Partition_Table] partition table. You can
217 create this with `gdisk /dev/sd(x)`. If there is no GPT, you cannot select the
220 If you want to use a separate DB/WAL device for your OSDs, you can specify it
221 through the '-journal_dev' option. The WAL is placed with the DB, if not
222 specified separately.
226 pveceph createosd /dev/sd[X] -journal_dev /dev/sd[Y]
229 NOTE: The DB stores BlueStore’s internal metadata and the WAL is BlueStore’s
230 internal journal or write-ahead log. It is recommended to use a fast SSD or
231 NVRAM for better performance.
236 Till Ceph luminous, Filestore was used as storage type for Ceph OSDs. It can
237 still be used and might give better performance in small setups, when backed by
238 an NVMe SSD or similar.
242 pveceph createosd /dev/sd[X] -bluestore 0
245 NOTE: In order to select a disk in the GUI, the disk needs to have a
246 GPT footnoteref:[GPT] partition table. You can
247 create this with `gdisk /dev/sd(x)`. If there is no GPT, you cannot select the
248 disk as journal. Currently the journal size is fixed to 5 GB.
250 If you want to use a dedicated SSD journal disk:
254 pveceph createosd /dev/sd[X] -journal_dev /dev/sd[Y] -bluestore 0
257 Example: Use /dev/sdf as data disk (4TB) and /dev/sdb is the dedicated SSD
262 pveceph createosd /dev/sdf -journal_dev /dev/sdb -bluestore 0
265 This partitions the disk (data and journal partition), creates
266 filesystems and starts the OSD, afterwards it is running and fully
269 NOTE: This command refuses to initialize disk when it detects existing data. So
270 if you want to overwrite a disk you should remove existing data first. You can
271 do that using: 'ceph-disk zap /dev/sd[X]'
273 You can create OSDs containing both journal and data partitions or you
274 can place the journal on a dedicated SSD. Using a SSD journal disk is
275 highly recommended to achieve good performance.
282 [thumbnail="screenshot/gui-ceph-pools.png"]
284 A pool is a logical group for storing objects. It holds **P**lacement
285 **G**roups (`PG`, `pg_num`), a collection of objects.
287 When no options are given, we set a default of **128 PGs**, a **size of 3
288 replicas** and a **min_size of 2 replicas** for serving objects in a degraded
291 NOTE: The default number of PGs works for 2-5 disks. Ceph throws a
292 'HEALTH_WARNING' if you have too few or too many PGs in your cluster.
294 It is advised to calculate the PG number depending on your setup, you can find
295 the formula and the PG calculator footnote:[PG calculator
296 http://ceph.com/pgcalc/] online. While PGs can be increased later on, they can
300 You can create pools through command line or on the GUI on each PVE host under
305 pveceph createpool <name>
308 If you would like to automatically get also a storage definition for your pool,
309 active the checkbox "Add storages" on the GUI or use the command line option
310 '--add_storages' on pool creation.
312 Further information on Ceph pool handling can be found in the Ceph pool
313 operation footnote:[Ceph pool operation
314 http://docs.ceph.com/docs/luminous/rados/operations/pools/]
317 Ceph CRUSH & device classes
318 ---------------------------
319 The foundation of Ceph is its algorithm, **C**ontrolled **R**eplication
320 **U**nder **S**calable **H**ashing
321 (CRUSH footnote:[CRUSH https://ceph.com/wp-content/uploads/2016/08/weil-crush-sc06.pdf]).
323 CRUSH calculates where to store to and retrieve data from, this has the
324 advantage that no central index service is needed. CRUSH works with a map of
325 OSDs, buckets (device locations) and rulesets (data replication) for pools.
327 NOTE: Further information can be found in the Ceph documentation, under the
328 section CRUSH map footnote:[CRUSH map http://docs.ceph.com/docs/luminous/rados/operations/crush-map/].
330 This map can be altered to reflect different replication hierarchies. The object
331 replicas can be separated (eg. failure domains), while maintaining the desired
334 A common use case is to use different classes of disks for different Ceph pools.
335 For this reason, Ceph introduced the device classes with luminous, to
336 accommodate the need for easy ruleset generation.
338 The device classes can be seen in the 'ceph osd tree' output. These classes
339 represent their own root bucket, which can be seen with the below command.
343 ceph osd crush tree --show-shadow
346 Example output form the above command:
350 ID CLASS WEIGHT TYPE NAME
351 -16 nvme 2.18307 root default~nvme
352 -13 nvme 0.72769 host sumi1~nvme
353 12 nvme 0.72769 osd.12
354 -14 nvme 0.72769 host sumi2~nvme
355 13 nvme 0.72769 osd.13
356 -15 nvme 0.72769 host sumi3~nvme
357 14 nvme 0.72769 osd.14
358 -1 7.70544 root default
359 -3 2.56848 host sumi1
360 12 nvme 0.72769 osd.12
361 -5 2.56848 host sumi2
362 13 nvme 0.72769 osd.13
363 -7 2.56848 host sumi3
364 14 nvme 0.72769 osd.14
367 To let a pool distribute its objects only on a specific device class, you need
368 to create a ruleset with the specific class first.
372 ceph osd crush rule create-replicated <rule-name> <root> <failure-domain> <class>
375 [frame="none",grid="none", align="left", cols="30%,70%"]
377 |<rule-name>|name of the rule, to connect with a pool (seen in GUI & CLI)
378 |<root>|which crush root it should belong to (default ceph root "default")
379 |<failure-domain>|at which failure-domain the objects should be distributed (usually host)
380 |<class>|what type of OSD backing store to use (eg. nvme, ssd, hdd)
383 Once the rule is in the CRUSH map, you can tell a pool to use the ruleset.
387 ceph osd pool set <pool-name> crush_rule <rule-name>
390 TIP: If the pool already contains objects, all of these have to be moved
391 accordingly. Depending on your setup this may introduce a big performance hit on
392 your cluster. As an alternative, you can create a new pool and move disks
399 [thumbnail="screenshot/gui-ceph-log.png"]
401 You can then configure {pve} to use such pools to store VM or
402 Container images. Simply use the GUI too add a new `RBD` storage (see
403 section xref:ceph_rados_block_devices[Ceph RADOS Block Devices (RBD)]).
405 You also need to copy the keyring to a predefined location for a external Ceph
406 cluster. If Ceph is installed on the Proxmox nodes itself, then this will be
409 NOTE: The file name needs to be `<storage_id> + `.keyring` - `<storage_id>` is
410 the expression after 'rbd:' in `/etc/pve/storage.cfg` which is
411 `my-ceph-storage` in the following example:
415 mkdir /etc/pve/priv/ceph
416 cp /etc/ceph/ceph.client.admin.keyring /etc/pve/priv/ceph/my-ceph-storage.keyring
423 Ceph provides also a filesystem running on top of the same object storage as
424 RADOS block devices do. A **M**eta**d**ata **S**erver (`MDS`) is used to map
425 the RADOS backed objects to files and directories, allowing to provide a
426 POSIX-compliant replicated filesystem. This allows one to have a clustered
427 highly available shared filesystem in an easy way if ceph is already used. Its
428 Metadata Servers guarantee that files get balanced out over the whole Ceph
429 cluster, this way even high load will not overload a single host, which can be
430 an issue with traditional shared filesystem approaches, like `NFS`, for
433 {pve} supports both, using an existing xref:storage_cephfs[CephFS as storage])
434 to save backups, ISO files or container templates and creating a
435 hyper-converged CephFS itself.
439 Metadata Server (MDS)
440 ~~~~~~~~~~~~~~~~~~~~~
442 CephFS needs at least one Metadata Server to be configured and running to be
443 able to work. One can simply create one through the {pve} web GUI's `Node ->
444 CephFS` panel or on the command line with:
450 Multiple metadata servers can be created in a cluster. But with the default
451 settings only one can be active at any time. If an MDS, or its node, becomes
452 unresponsive (or crashes), another `standby` MDS will get promoted to `active`.
453 One can speed up the hand-over between the active and a standby MDS up by using
454 the 'hotstandby' parameter option on create, or if you have already created it
458 mds standby replay = true
461 in the ceph.conf respective MDS section. With this enabled, this specific MDS
462 will always poll the active one, so that it can take over faster as it is in a
463 `warm` state. But naturally, the active polling will cause some additional
464 performance impact on your system and active `MDS`.
469 Since Luminous (12.2.x) you can also have multiple active metadata servers
470 running, but this is normally only useful for a high count on parallel clients,
471 as else the `MDS` seldom is the bottleneck. If you want to set this up please
472 refer to the ceph documentation. footnote:[Configuring multiple active MDS
473 daemons http://docs.ceph.com/docs/luminous/cephfs/multimds/]
475 [[pveceph_fs_create]]
479 With {pve}'s CephFS integration into you can create a CephFS easily over the
480 Web GUI, the CLI or an external API interface. Some prerequisites are required
483 .Prerequisites for a successful CephFS setup:
484 - xref:pve_ceph_install[Install Ceph packages], if this was already done some
485 time ago you might want to rerun it on an up to date system to ensure that
486 also all CephFS related packages get installed.
487 - xref:pve_ceph_monitors[Setup Monitors]
488 - xref:pve_ceph_monitors[Setup your OSDs]
489 - xref:pveceph_fs_mds[Setup at least one MDS]
491 After this got all checked and done you can simply create a CephFS through
492 either the Web GUI's `Node -> CephFS` panel or the command line tool `pveceph`,
496 pveceph fs create --pg_num 128 --add-storage
499 This creates a CephFS named `'cephfs'' using a pool for its data named
500 `'cephfs_data'' with `128` placement groups and a pool for its metadata named
501 `'cephfs_metadata'' with one quarter of the data pools placement groups (`32`).
502 Check the xref:pve_ceph_pools[{pve} managed Ceph pool chapter] or visit the
503 Ceph documentation for more information regarding a fitting placement group
504 number (`pg_num`) for your setup footnote:[Ceph Placement Groups
505 http://docs.ceph.com/docs/luminous/rados/operations/placement-groups/].
506 Additionally, the `'--add-storage'' parameter will add the CephFS to the {pve}
507 storage configuration after it was created successfully.
512 WARNING: Destroying a CephFS will render all its data unusable, this cannot be
515 If you really want to destroy an existing CephFS you first need to stop, or
516 destroy, all metadata server (`M̀DS`). You can destroy them either over the Web
517 GUI or the command line interface, with:
520 pveceph mds destroy NAME
522 on each {pve} node hosting a MDS daemon.
524 Then, you can remove (destroy) CephFS by issuing a:
527 ceph fs rm NAME --yes-i-really-mean-it
529 on a single node hosting Ceph. After this you may want to remove the created
530 data and metadata pools, this can be done either over the Web GUI or the CLI
534 pveceph pool destroy NAME
538 include::pve-copyright.adoc[]