]>
Commit | Line | Data |
---|---|---|
1 | [[chapter_pveceph]] | |
2 | ifdef::manvolnum[] | |
3 | pveceph(1) | |
4 | ========== | |
5 | :pve-toplevel: | |
6 | ||
7 | NAME | |
8 | ---- | |
9 | ||
10 | pveceph - Manage Ceph Services on Proxmox VE Nodes | |
11 | ||
12 | SYNOPSIS | |
13 | -------- | |
14 | ||
15 | include::pveceph.1-synopsis.adoc[] | |
16 | ||
17 | DESCRIPTION | |
18 | ----------- | |
19 | endif::manvolnum[] | |
20 | ifndef::manvolnum[] | |
21 | Manage Ceph Services on Proxmox VE Nodes | |
22 | ======================================== | |
23 | :pve-toplevel: | |
24 | endif::manvolnum[] | |
25 | ||
26 | [thumbnail="gui-ceph-status.png"] | |
27 | ||
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. | |
36 | ||
37 | Ceph is a distributed object store and file system designed to provide | |
38 | excellent performance, reliability and scalability. | |
39 | ||
40 | .Some of the advantages of Ceph are: | |
41 | - Easy setup and management with CLI and GUI support on Proxmox VE | |
42 | - Thin provisioning | |
43 | - Snapshots support | |
44 | - Self healing | |
45 | - No single point of failure | |
46 | - Scalable to the exabyte level | |
47 | - Setup pools with different performance and redundancy characteristics | |
48 | - Data is replicated, making it fault tolerant | |
49 | - Runs on economical commodity hardware | |
50 | - No need for hardware RAID controllers | |
51 | - Easy management | |
52 | - Open source | |
53 | ||
54 | For small to mid sized deployments, it is possible to install a Ceph server for | |
55 | RADOS Block Devices (RBD) directly on your {pve} cluster nodes, see | |
56 | xref:ceph_rados_block_devices[Ceph RADOS Block Devices (RBD)]. Recent | |
57 | hardware has plenty of CPU power and RAM, so running storage services | |
58 | and VMs on the same node is possible. | |
59 | ||
60 | To simplify management, we provide 'pveceph' - a tool to install and | |
61 | manage {ceph} services on {pve} nodes. | |
62 | ||
63 | .Ceph consists of a couple of Daemons footnote:[Ceph intro http://docs.ceph.com/docs/master/start/intro/], for use as a RBD storage: | |
64 | - Ceph Monitor (ceph-mon) | |
65 | - Ceph Manager (ceph-mgr) | |
66 | - Ceph OSD (ceph-osd; Object Storage Daemon) | |
67 | ||
68 | TIP: We recommend to get familiar with the Ceph vocabulary. | |
69 | footnote:[Ceph glossary http://docs.ceph.com/docs/luminous/glossary] | |
70 | ||
71 | ||
72 | Precondition | |
73 | ------------ | |
74 | ||
75 | To build a Proxmox Ceph Cluster there should be at least three (preferably) | |
76 | identical servers for the setup. | |
77 | ||
78 | A 10Gb network, exclusively used for Ceph, is recommended. A meshed network | |
79 | setup is also an option if there are no 10Gb switches available, see our wiki | |
80 | article footnote:[Full Mesh Network for Ceph {webwiki-url}Full_Mesh_Network_for_Ceph_Server] . | |
81 | ||
82 | Check also the recommendations from | |
83 | http://docs.ceph.com/docs/luminous/start/hardware-recommendations/[Ceph's website]. | |
84 | ||
85 | .Avoid RAID | |
86 | While RAID controller are build for storage virtualisation, to combine | |
87 | independent disks to form one or more logical units. Their caching methods, | |
88 | algorithms (RAID modes; incl. JBOD), disk or write/read optimisations are | |
89 | targeted towards aforementioned logical units and not to Ceph. | |
90 | ||
91 | WARNING: Avoid RAID controller, use host bus adapter (HBA) instead. | |
92 | ||
93 | ||
94 | Installation of Ceph Packages | |
95 | ----------------------------- | |
96 | ||
97 | On each node run the installation script as follows: | |
98 | ||
99 | [source,bash] | |
100 | ---- | |
101 | pveceph install | |
102 | ---- | |
103 | ||
104 | This sets up an `apt` package repository in | |
105 | `/etc/apt/sources.list.d/ceph.list` and installs the required software. | |
106 | ||
107 | ||
108 | Creating initial Ceph configuration | |
109 | ----------------------------------- | |
110 | ||
111 | [thumbnail="gui-ceph-config.png"] | |
112 | ||
113 | After installation of packages, you need to create an initial Ceph | |
114 | configuration on just one node, based on your network (`10.10.10.0/24` | |
115 | in the following example) dedicated for Ceph: | |
116 | ||
117 | [source,bash] | |
118 | ---- | |
119 | pveceph init --network 10.10.10.0/24 | |
120 | ---- | |
121 | ||
122 | This creates an initial configuration at `/etc/pve/ceph.conf`. That file is | |
123 | automatically distributed to all {pve} nodes by using | |
124 | xref:chapter_pmxcfs[pmxcfs]. The command also creates a symbolic link | |
125 | from `/etc/ceph/ceph.conf` pointing to that file. So you can simply run | |
126 | Ceph commands without the need to specify a configuration file. | |
127 | ||
128 | ||
129 | [[pve_ceph_monitors]] | |
130 | Creating Ceph Monitors | |
131 | ---------------------- | |
132 | ||
133 | [thumbnail="gui-ceph-monitor.png"] | |
134 | ||
135 | The Ceph Monitor (MON) | |
136 | footnote:[Ceph Monitor http://docs.ceph.com/docs/luminous/start/intro/] | |
137 | maintains a master copy of the cluster map. For high availability you need to | |
138 | have at least 3 monitors. | |
139 | ||
140 | On each node where you want to place a monitor (three monitors are recommended), | |
141 | create it by using the 'Ceph -> Monitor' tab in the GUI or run. | |
142 | ||
143 | ||
144 | [source,bash] | |
145 | ---- | |
146 | pveceph createmon | |
147 | ---- | |
148 | ||
149 | This will also install the needed Ceph Manager ('ceph-mgr') by default. If you | |
150 | do not want to install a manager, specify the '-exclude-manager' option. | |
151 | ||
152 | ||
153 | [[pve_ceph_manager]] | |
154 | Creating Ceph Manager | |
155 | ---------------------- | |
156 | ||
157 | The Manager daemon runs alongside the monitors, providing an interface for | |
158 | monitoring the cluster. Since the Ceph luminous release the | |
159 | ceph-mgr footnote:[Ceph Manager http://docs.ceph.com/docs/luminous/mgr/] daemon | |
160 | is required. During monitor installation the ceph manager will be installed as | |
161 | well. | |
162 | ||
163 | NOTE: It is recommended to install the Ceph Manager on the monitor nodes. For | |
164 | high availability install more then one manager. | |
165 | ||
166 | [source,bash] | |
167 | ---- | |
168 | pveceph createmgr | |
169 | ---- | |
170 | ||
171 | ||
172 | [[pve_ceph_osds]] | |
173 | Creating Ceph OSDs | |
174 | ------------------ | |
175 | ||
176 | [thumbnail="gui-ceph-osd-status.png"] | |
177 | ||
178 | via GUI or via CLI as follows: | |
179 | ||
180 | [source,bash] | |
181 | ---- | |
182 | pveceph createosd /dev/sd[X] | |
183 | ---- | |
184 | ||
185 | TIP: We recommend a Ceph cluster size, starting with 12 OSDs, distributed evenly | |
186 | among your, at least three nodes (4 OSDs on each node). | |
187 | ||
188 | If the disk was used before (eg. ZFS/RAID/OSD), to remove partition table, boot | |
189 | sector and any OSD leftover the following commands should be sufficient. | |
190 | ||
191 | [source,bash] | |
192 | ---- | |
193 | dd if=/dev/zero of=/dev/sd[X] bs=1M count=200 | |
194 | ceph-disk zap /dev/sd[X] | |
195 | ---- | |
196 | ||
197 | WARNING: The above commands will destroy data on the disk! | |
198 | ||
199 | Ceph Bluestore | |
200 | ~~~~~~~~~~~~~~ | |
201 | ||
202 | Starting with the Ceph Kraken release, a new Ceph OSD storage type was | |
203 | introduced, the so called Bluestore | |
204 | footnote:[Ceph Bluestore http://ceph.com/community/new-luminous-bluestore/]. | |
205 | This is the default when creating OSDs in Ceph luminous. | |
206 | ||
207 | [source,bash] | |
208 | ---- | |
209 | pveceph createosd /dev/sd[X] | |
210 | ---- | |
211 | ||
212 | NOTE: In order to select a disk in the GUI, to be more failsafe, the disk needs | |
213 | to have a GPT footnoteref:[GPT, GPT partition table | |
214 | https://en.wikipedia.org/wiki/GUID_Partition_Table] partition table. You can | |
215 | create this with `gdisk /dev/sd(x)`. If there is no GPT, you cannot select the | |
216 | disk as DB/WAL. | |
217 | ||
218 | If you want to use a separate DB/WAL device for your OSDs, you can specify it | |
219 | through the '-journal_dev' option. The WAL is placed with the DB, if not | |
220 | specified separately. | |
221 | ||
222 | [source,bash] | |
223 | ---- | |
224 | pveceph createosd /dev/sd[X] -journal_dev /dev/sd[Y] | |
225 | ---- | |
226 | ||
227 | NOTE: The DB stores BlueStore’s internal metadata and the WAL is BlueStore’s | |
228 | internal journal or write-ahead log. It is recommended to use a fast SSDs or | |
229 | NVRAM for better performance. | |
230 | ||
231 | ||
232 | Ceph Filestore | |
233 | ~~~~~~~~~~~~~ | |
234 | Till Ceph luminous, Filestore was used as storage type for Ceph OSDs. It can | |
235 | still be used and might give better performance in small setups, when backed by | |
236 | a NVMe SSD or similar. | |
237 | ||
238 | [source,bash] | |
239 | ---- | |
240 | pveceph createosd /dev/sd[X] -bluestore 0 | |
241 | ---- | |
242 | ||
243 | NOTE: In order to select a disk in the GUI, the disk needs to have a | |
244 | GPT footnoteref:[GPT] partition table. You can | |
245 | create this with `gdisk /dev/sd(x)`. If there is no GPT, you cannot select the | |
246 | disk as journal. Currently the journal size is fixed to 5 GB. | |
247 | ||
248 | If you want to use a dedicated SSD journal disk: | |
249 | ||
250 | [source,bash] | |
251 | ---- | |
252 | pveceph createosd /dev/sd[X] -journal_dev /dev/sd[Y] -bluestore 0 | |
253 | ---- | |
254 | ||
255 | Example: Use /dev/sdf as data disk (4TB) and /dev/sdb is the dedicated SSD | |
256 | journal disk. | |
257 | ||
258 | [source,bash] | |
259 | ---- | |
260 | pveceph createosd /dev/sdf -journal_dev /dev/sdb -bluestore 0 | |
261 | ---- | |
262 | ||
263 | This partitions the disk (data and journal partition), creates | |
264 | filesystems and starts the OSD, afterwards it is running and fully | |
265 | functional. | |
266 | ||
267 | NOTE: This command refuses to initialize disk when it detects existing data. So | |
268 | if you want to overwrite a disk you should remove existing data first. You can | |
269 | do that using: 'ceph-disk zap /dev/sd[X]' | |
270 | ||
271 | You can create OSDs containing both journal and data partitions or you | |
272 | can place the journal on a dedicated SSD. Using a SSD journal disk is | |
273 | highly recommended to achieve good performance. | |
274 | ||
275 | ||
276 | [[pve_ceph_pools]] | |
277 | Creating Ceph Pools | |
278 | ------------------- | |
279 | ||
280 | [thumbnail="gui-ceph-pools.png"] | |
281 | ||
282 | A pool is a logical group for storing objects. It holds **P**lacement | |
283 | **G**roups (PG), a collection of objects. | |
284 | ||
285 | When no options are given, we set a | |
286 | default of **64 PGs**, a **size of 3 replicas** and a **min_size of 2 replicas** | |
287 | for serving objects in a degraded state. | |
288 | ||
289 | NOTE: The default number of PGs works for 2-6 disks. Ceph throws a | |
290 | "HEALTH_WARNING" if you have too few or too many PGs in your cluster. | |
291 | ||
292 | It is advised to calculate the PG number depending on your setup, you can find | |
293 | the formula and the PG calculator footnote:[PG calculator | |
294 | http://ceph.com/pgcalc/] online. While PGs can be increased later on, they can | |
295 | never be decreased. | |
296 | ||
297 | ||
298 | You can create pools through command line or on the GUI on each PVE host under | |
299 | **Ceph -> Pools**. | |
300 | ||
301 | [source,bash] | |
302 | ---- | |
303 | pveceph createpool <name> | |
304 | ---- | |
305 | ||
306 | If you would like to automatically get also a storage definition for your pool, | |
307 | active the checkbox "Add storages" on the GUI or use the command line option | |
308 | '--add_storages' on pool creation. | |
309 | ||
310 | Further information on Ceph pool handling can be found in the Ceph pool | |
311 | operation footnote:[Ceph pool operation | |
312 | http://docs.ceph.com/docs/luminous/rados/operations/pools/] | |
313 | manual. | |
314 | ||
315 | Ceph CRUSH & device classes | |
316 | --------------------------- | |
317 | The foundation of Ceph is its algorithm, **C**ontrolled **R**eplication | |
318 | **U**nder **S**calable **H**ashing | |
319 | (CRUSH footnote:[CRUSH https://ceph.com/wp-content/uploads/2016/08/weil-crush-sc06.pdf]). | |
320 | ||
321 | CRUSH calculates where to store to and retrieve data from, this has the | |
322 | advantage that no central index service is needed. CRUSH works with a map of | |
323 | OSDs, buckets (device locations) and rulesets (data replication) for pools. | |
324 | ||
325 | NOTE: Further information can be found in the Ceph documentation, under the | |
326 | section CRUSH map footnote:[CRUSH map http://docs.ceph.com/docs/luminous/rados/operations/crush-map/]. | |
327 | ||
328 | This map can be altered to reflect different replication hierarchies. The object | |
329 | replicas can be separated (eg. failure domains), while maintaining the desired | |
330 | distribution. | |
331 | ||
332 | A common use case is to use different classes of disks for different Ceph pools. | |
333 | For this reason, Ceph introduced the device classes with luminous, to | |
334 | accommodate the need for easy ruleset generation. | |
335 | ||
336 | The device classes can be seen in the 'ceph osd tree' output. These classes | |
337 | represent their own root bucket, which can be seen with the below command. | |
338 | ||
339 | [source, bash] | |
340 | ---- | |
341 | ceph osd crush tree --show-shadow | |
342 | ---- | |
343 | ||
344 | Example output form the above command: | |
345 | ||
346 | [source, bash] | |
347 | ---- | |
348 | ID CLASS WEIGHT TYPE NAME | |
349 | -16 nvme 2.18307 root default~nvme | |
350 | -13 nvme 0.72769 host sumi1~nvme | |
351 | 12 nvme 0.72769 osd.12 | |
352 | -14 nvme 0.72769 host sumi2~nvme | |
353 | 13 nvme 0.72769 osd.13 | |
354 | -15 nvme 0.72769 host sumi3~nvme | |
355 | 14 nvme 0.72769 osd.14 | |
356 | -1 7.70544 root default | |
357 | -3 2.56848 host sumi1 | |
358 | 12 nvme 0.72769 osd.12 | |
359 | -5 2.56848 host sumi2 | |
360 | 13 nvme 0.72769 osd.13 | |
361 | -7 2.56848 host sumi3 | |
362 | 14 nvme 0.72769 osd.14 | |
363 | ---- | |
364 | ||
365 | To let a pool distribute its objects only on a specific device class, you need | |
366 | to create a ruleset with the specific class first. | |
367 | ||
368 | [source, bash] | |
369 | ---- | |
370 | ceph osd crush rule create-replicated <rule-name> <root> <failure-domain> <class> | |
371 | ---- | |
372 | ||
373 | [frame="none",grid="none", align="left", cols="30%,70%"] | |
374 | |=== | |
375 | |<rule-name>|name of the rule, to connect with a pool (seen in GUI & CLI) | |
376 | |<root>|which crush root it should belong to (default ceph root "default") | |
377 | |<failure-domain>|at which failure-domain the objects should be distributed (usually host) | |
378 | |<class>|what type of OSD backing store to use (eg. nvme, ssd, hdd) | |
379 | |=== | |
380 | ||
381 | Once the rule is in the CRUSH map, you can tell a pool to use the ruleset. | |
382 | ||
383 | [source, bash] | |
384 | ---- | |
385 | ceph osd pool set <pool-name> crush_rule <rule-name> | |
386 | ---- | |
387 | ||
388 | TIP: If the pool already contains objects, all of these have to be moved | |
389 | accordingly. Depending on your setup this may introduce a big performance hit on | |
390 | your cluster. As an alternative, you can create a new pool and move disks | |
391 | separately. | |
392 | ||
393 | ||
394 | Ceph Client | |
395 | ----------- | |
396 | ||
397 | [thumbnail="gui-ceph-log.png"] | |
398 | ||
399 | You can then configure {pve} to use such pools to store VM or | |
400 | Container images. Simply use the GUI too add a new `RBD` storage (see | |
401 | section xref:ceph_rados_block_devices[Ceph RADOS Block Devices (RBD)]). | |
402 | ||
403 | You also need to copy the keyring to a predefined location for a external Ceph | |
404 | cluster. If Ceph is installed on the Proxmox nodes itself, then this will be | |
405 | done automatically. | |
406 | ||
407 | NOTE: The file name needs to be `<storage_id> + `.keyring` - `<storage_id>` is | |
408 | the expression after 'rbd:' in `/etc/pve/storage.cfg` which is | |
409 | `my-ceph-storage` in the following example: | |
410 | ||
411 | [source,bash] | |
412 | ---- | |
413 | mkdir /etc/pve/priv/ceph | |
414 | cp /etc/ceph/ceph.client.admin.keyring /etc/pve/priv/ceph/my-ceph-storage.keyring | |
415 | ---- | |
416 | ||
417 | ||
418 | ifdef::manvolnum[] | |
419 | include::pve-copyright.adoc[] | |
420 | endif::manvolnum[] |