describe begin/end markers and .pve-ignore.
[pve-docs.git] / pct.adoc
CommitLineData
0c6b782f
DM
1ifdef::manvolnum[]
2PVE({manvolnum})
3================
38fd0958 4include::attributes.txt[]
0c6b782f
DM
5
6NAME
7----
8
9pct - Tool to manage Linux Containers (LXC) on Proxmox VE
10
11
12SYNOPSYS
13--------
14
15include::pct.1-synopsis.adoc[]
16
17DESCRIPTION
18-----------
19endif::manvolnum[]
20
21ifndef::manvolnum[]
22Proxmox Container Toolkit
23=========================
38fd0958 24include::attributes.txt[]
0c6b782f
DM
25endif::manvolnum[]
26
4a2ae9ed
DM
27
28Containers are a lightweight alternative to fully virtualized
29VMs. Instead of emulating a complete Operating System (OS), containers
30simply use the OS of the host they run on. This implies that all
31containers use the same kernel, and that they can access resources
32from the host directly.
33
34This is great because containers do not waste CPU power nor memory due
35to kernel emulation. Container run-time costs are close to zero and
36usually negligible. But there are also some drawbacks you need to
37consider:
38
39* You can only run Linux based OS inside containers, i.e. it is not
a8e99754 40 possible to run FreeBSD or MS Windows inside.
4a2ae9ed 41
a8e99754 42* For security reasons, access to host resources needs to be
4a2ae9ed 43 restricted. This is done with AppArmor, SecComp filters and other
a8e99754 44 kernel features. Be prepared that some syscalls are not allowed
4a2ae9ed
DM
45 inside containers.
46
47{pve} uses https://linuxcontainers.org/[LXC] as underlying container
48technology. We consider LXC as low-level library, which provides
a8e99754 49countless options. It would be too difficult to use those tools
4a2ae9ed
DM
50directly. Instead, we provide a small wrapper called `pct`, the
51"Proxmox Container Toolkit".
52
a8e99754 53The toolkit is tightly coupled with {pve}. That means that it is aware
4a2ae9ed
DM
54of the cluster setup, and it can use the same network and storage
55resources as fully virtualized VMs. You can even use the {pve}
56firewall, or manage containers using the HA framework.
57
58Our primary goal is to offer an environment as one would get from a
59VM, but without the additional overhead. We call this "System
60Containers".
61
70a42028
DM
62NOTE: If you want to run micro-containers (with docker, rct, ...), it
63is best to run them inside a VM.
4a2ae9ed
DM
64
65
66Security Considerations
67-----------------------
68
69Containers use the same kernel as the host, so there is a big attack
70surface for malicious users. You should consider this fact if you
71provide containers to totally untrusted people. In general, fully
a8e99754 72virtualized VMs provide better isolation.
4a2ae9ed
DM
73
74The good news is that LXC uses many kernel security features like
75AppArmor, CGroups and PID and user namespaces, which makes containers
76usage quite secure. We distinguish two types of containers:
77
78Privileged containers
79~~~~~~~~~~~~~~~~~~~~~
80
81Security is done by dropping capabilities, using mandatory access
82control (AppArmor), SecComp filters and namespaces. The LXC team
83considers this kind of container as unsafe, and they will not consider
84new container escape exploits to be security issues worthy of a CVE
85and quick fix. So you should use this kind of containers only inside a
86trusted environment, or when no untrusted task is running as root in
87the container.
88
89Unprivileged containers
90~~~~~~~~~~~~~~~~~~~~~~~
91
a8e99754 92This kind of containers use a new kernel feature called user
4a2ae9ed
DM
93namespaces. The root uid 0 inside the container is mapped to an
94unprivileged user outside the container. This means that most security
95issues (container escape, resource abuse, ...) in those containers
96will affect a random unprivileged user, and so would be a generic
a8e99754 97kernel security bug rather than an LXC issue. The LXC team thinks
4a2ae9ed
DM
98unprivileged containers are safe by design.
99
7fc230db
DM
100
101Configuration
102-------------
103
166e63d6
FG
104The '/etc/pve/lxc/<CTID>.conf' file stores container configuration,
105where '<CTID>' is the numeric ID of the given container. Like all
106other files stored inside '/etc/pve/', they get automatically
107replicated to all other cluster nodes.
108
109NOTE: CTIDs < 100 are reserved for internal purposes, and CTIDs need to be
110unique cluster wide.
7fc230db 111
105bc8f1
DM
112.Example Container Configuration
113----
114ostype: debian
115arch: amd64
116hostname: www
117memory: 512
118swap: 512
119net0: bridge=vmbr0,hwaddr=66:64:66:64:64:36,ip=dhcp,name=eth0,type=veth
120rootfs: local:107/vm-107-disk-1.raw,size=7G
121----
122
7fc230db 123Those configuration files are simple text files, and you can edit them
55fb2a21
DM
124using a normal text editor ('vi', 'nano', ...). This is sometimes
125useful to do small corrections, but keep in mind that you need to
126restart the container to apply such changes.
127
128For that reason, it is usually better to use the 'pct' command to
129generate and modify those files, or do the whole thing using the GUI.
130Our toolkit is smart enough to instantaneously apply most changes to
105bc8f1
DM
131running containers. This feature is called "hot plug", and there is no
132need to restart the container in that case.
7fc230db
DM
133
134File Format
135~~~~~~~~~~~
136
137Container configuration files use a simple colon separated key/value
138format. Each line has the following format:
139
140 # this is a comment
141 OPTION: value
142
143Blank lines in those files are ignored, and lines starting with a '#'
144character are treated as comments and are also ignored.
145
146It is possible to add low-level, LXC style configuration directly, for
147example:
148
149 lxc.init_cmd: /sbin/my_own_init
150
151or
152
153 lxc.init_cmd = /sbin/my_own_init
154
155Those settings are directly passed to the LXC low-level tools.
156
105bc8f1
DM
157Snapshots
158~~~~~~~~~
159
160When you create a snapshot, 'pct' stores the configuration at snapshot
161time into a separate snapshot section within the same configuration
162file. For example, after creating a snapshot called 'testsnapshot',
163your configuration file will look like this:
164
165.Container Configuration with Snapshot
166----
167memory: 512
168swap: 512
169parent: testsnaphot
170...
171
172[testsnaphot]
173memory: 512
174swap: 512
175snaptime: 1457170803
176...
177----
178
a8e99754
FG
179There are a few snapshot related properties like 'parent' and
180'snaptime'. The 'parent' property is used to store the parent/child
105bc8f1
DM
181relationship between snapshots. 'snaptime' is the snapshot creation
182time stamp (unix epoch).
7fc230db 183
3f13c1c3
DM
184Guest Operating System Configuration
185~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
186
187We normally try to detect the operating system type inside the
188container, and then modify some files inside the container to make
189them work as expected. Here is a short list of things we do at
190container startup:
191
192set /etc/hostname:: to set the container name
193
a8e99754 194modify /etc/hosts:: to allow lookup of the local hostname
3f13c1c3
DM
195
196network setup:: pass the complete network setup to the container
197
198configure DNS:: pass information about DNS servers
199
a8e99754 200adapt the init system:: for example, fix the number of spawned getty processes
3f13c1c3
DM
201
202set the root password:: when creating a new container
203
204rewrite ssh_host_keys:: so that each container has unique keys
205
a8e99754 206randomize crontab:: so that cron does not start at the same time on all containers
3f13c1c3 207
25535d34
WB
208Changes made by {PVE} are enclosed by comment markers:
209
210 # --- BEGIN PVE ---
211 <data>
212 # --- END PVE ---
213
214If no such section is found it will be inserted at a reasonable location
215in the file.
216
217If such a section already exists it will be updated in place and will not be
218moved.
219
220Modification of a file can be prevented by adding a `.pve-ignore.` file for it.
221For instance, if the file `/etc/.pve-ignore.hosts` exists then the
222`/etc/hosts` file will not be touched. (This can be a simple empty file creatd
223via:
224
225 # touch /etc/.pve-ignore.hosts
226
227The above tasks are OS dependent and so they differ between different
228distributions. You can also disable all modifications entirely by manually
3f13c1c3
DM
229setting the 'ostype' to 'unmanaged'.
230
231OS type detection is done by testing for certain files inside the
232container:
233
234Ubuntu:: inspect /etc/lsb-release ('DISTRIB_ID=Ubuntu')
235
236Debian:: test /etc/debian_version
237
238Fedora:: test /etc/fedora-release
239
240RedHat or CentOS:: test /etc/redhat-release
241
242ArchLinux:: test /etc/arch-release
243
244Alpine:: test /etc/alpine-release
245
a8e99754 246NOTE: Container start fails if the configured 'ostype' differs from the auto
3f13c1c3
DM
247detected type.
248
d61bab51
DM
249
250Container Images
251----------------
252
a8e99754
FG
253Container Images, sometimes also referred to as "templates" or
254"appliances", are 'tar' archives which contain everything to run a
d61bab51
DM
255container. You can think of it as a tidy container backup. Like most
256modern container toolkits, 'pct' uses those images when you create a
257new container, for example:
258
259 pct create 999 local:vztmpl/debian-8.0-standard_8.0-1_amd64.tar.gz
260
261Proxmox itself ships a set of basic templates for most common
262operating systems, and you can download them using the 'pveam' (short
263for {pve} Appliance Manager) command line utility. You can also
264download https://www.turnkeylinux.org/[TurnKey Linux] containers using
265that tool (or the graphical user interface).
266
3a6fa247
DM
267Our image repositories contain a list of available images, and there
268is a cron job run each day to download that list. You can trigger that
269update manually with:
270
271 pveam update
272
273After that you can view the list of available images using:
274
275 pveam available
276
277You can restrict this large list by specifying the 'section' you are
278interested in, for example basic 'system' images:
279
280.List available system images
281----
282# pveam available --section system
283system archlinux-base_2015-24-29-1_x86_64.tar.gz
284system centos-7-default_20160205_amd64.tar.xz
285system debian-6.0-standard_6.0-7_amd64.tar.gz
286system debian-7.0-standard_7.0-3_amd64.tar.gz
287system debian-8.0-standard_8.0-1_amd64.tar.gz
288system ubuntu-12.04-standard_12.04-1_amd64.tar.gz
289system ubuntu-14.04-standard_14.04-1_amd64.tar.gz
290system ubuntu-15.04-standard_15.04-1_amd64.tar.gz
291system ubuntu-15.10-standard_15.10-1_amd64.tar.gz
292----
293
a8e99754 294Before you can use such a template, you need to download them into one
3a6fa247
DM
295of your storages. You can simply use storage 'local' for that
296purpose. For clustered installations, it is preferred to use a shared
297storage so that all nodes can access those images.
298
299 pveam download local debian-8.0-standard_8.0-1_amd64.tar.gz
300
24f73a63
DM
301You are now ready to create containers using that image, and you can
302list all downloaded images on storage 'local' with:
303
304----
305# pveam list local
306local:vztmpl/debian-8.0-standard_8.0-1_amd64.tar.gz 190.20MB
307----
308
a8e99754 309The above command shows you the full {pve} volume identifiers. They include
24f73a63
DM
310the storage name, and most other {pve} commands can use them. For
311examply you can delete that image later with:
312
313 pveam remove local:vztmpl/debian-8.0-standard_8.0-1_amd64.tar.gz
3a6fa247 314
d61bab51 315
70a42028
DM
316Container Storage
317-----------------
318
319Traditional containers use a very simple storage model, only allowing
320a single mount point, the root file system. This was further
321restricted to specific file system types like 'ext4' and 'nfs'.
322Additional mounts are often done by user provided scripts. This turend
a8e99754 323out to be complex and error prone, so we try to avoid that now.
70a42028
DM
324
325Our new LXC based container model is more flexible regarding
326storage. First, you can have more than a single mount point. This
327allows you to choose a suitable storage for each application. For
328example, you can use a relatively slow (and thus cheap) storage for
329the container root file system. Then you can use a second mount point
330to mount a very fast, distributed storage for your database
331application.
332
333The second big improvement is that you can use any storage type
334supported by the {pve} storage library. That means that you can store
335your containers on local 'lvmthin' or 'zfs', shared 'iSCSI' storage,
a8e99754 336or even on distributed storage systems like 'ceph'. It also enables us
70a42028 337to use advanced storage features like snapshots and clones. 'vzdump'
a8e99754 338can also use the snapshot feature to provide consistent container
70a42028
DM
339backups.
340
341Last but not least, you can also mount local devices directly, or
342mount local directories using bind mounts. That way you can access
343local storage inside containers with zero overhead. Such bind mounts
a8e99754 344also provide an easy way to share data between different containers.
70a42028 345
eeecce95 346
9e44e493
DM
347Mount Points
348~~~~~~~~~~~~
eeecce95 349
9e44e493
DM
350Beside the root directory the container can also have additional mount points.
351Currently there are basically three types of mount points: storage backed
352mount points, bind mounts and device mounts.
353
354Storage backed mount points are managed by the {pve} storage subsystem and come
eeecce95
WB
355in three different flavors:
356
357- Image based: These are raw images containing a single ext4 formatted file
358 system.
359- ZFS Subvolumes: These are technically bind mounts, but with managed storage,
360 and thus allow resizing and snapshotting.
361- Directories: passing `size=0` triggers a special case where instead of a raw
362 image a directory is created.
363
364Bind mounts are considered to not be managed by the storage subsystem, so you
365cannot make snapshots or deal with quotas from inside the container, and with
366unprivileged containers you might run into permission problems caused by the
367user mapping, and cannot use ACLs from inside an unprivileged container.
368
369Similarly device mounts are not managed by the storage, but for these the
370`quota` and `acl` options will be honored.
371
22a74065
FG
372WARNING: Because of existing issues in the Linux kernel's freezer
373subsystem the usage of FUSE mounts inside a container is strongly
374advised against, as containers need to be frozen for suspend or
375snapshot mode backups. If FUSE mounts cannot be replaced by other
376mounting mechanisms or storage technologies, it is possible to
377establish the FUSE mount on the Proxmox host and use a bind
9e44e493 378mount point to make it accessible inside the container.
eeecce95 379
9e44e493 380The root mount point is configured with the 'rootfs' property, and you can
fe154a4f
DM
381configure up to 10 additional mount points. The corresponding options
382are called 'mp0' to 'mp9', and they can contain the following setting:
383
384include::pct-mountpoint-opts.adoc[]
385
386.Typical Container 'rootfs' configuration
387----
388rootfs: thin1:base-100-disk-1,size=8G
389----
390
d6ed3622 391Using quotas inside containers
04c569f6 392~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
d6ed3622 393
9e44e493
DM
394Quotas allow to set limits inside a container for the amount of disk
395space that each user can use. This only works on ext4 image based
396storage types and currently does not work with unprivileged
397containers.
d6ed3622 398
9e44e493
DM
399Activating the `quota` option causes the following mount options to be
400used for a mount point:
401`usrjquota=aquota.user,grpjquota=aquota.group,jqfmt=vfsv0`
d6ed3622 402
9e44e493
DM
403This allows quotas to be used like you would on any other system. You
404can initialize the `/aquota.user` and `/aquota.group` files by running
d6ed3622 405
9e44e493
DM
406----
407quotacheck -cmug /
408quotaon /
409----
d6ed3622 410
166e63d6
FG
411and edit the quotas via the `edquota` command. Refer to the documentation
412of the distribution running inside the container for details.
413
9e44e493
DM
414NOTE: You need to run the above commands for every mount point by passing
415the mount point's path instead of just `/`.
416
d6ed3622 417
6c60aebf 418Using ACLs inside containers
04c569f6 419~~~~~~~~~~~~~~~~~~~~~~~~~~~~
6c60aebf
EK
420
421The standard Posix Access Control Lists are also available inside containers.
422ACLs allow you to set more detailed file ownership than the traditional user/
423group/others model.
d6ed3622 424
04c569f6
DM
425
426Container Network
427-----------------
428
bac8c385
DM
429You can configure up to 10 network interfaces for a single
430container. The corresponding options are called 'net0' to 'net9', and
431they can contain the following setting:
432
433include::pct-network-opts.adoc[]
04c569f6
DM
434
435
436Managing Containers with 'pct'
437------------------------------
438
439'pct' is the tool to manage Linux Containers on {pve}. You can create
440and destroy containers, and control execution (start, stop, migrate,
441...). You can use pct to set parameters in the associated config file,
442like network configuration or memory limits.
443
444CLI Usage Examples
445~~~~~~~~~~~~~~~~~~
446
447Create a container based on a Debian template (provided you have
448already downloaded the template via the webgui)
449
450 pct create 100 /var/lib/vz/template/cache/debian-8.0-standard_8.0-1_amd64.tar.gz
451
452Start container 100
453
454 pct start 100
455
456Start a login session via getty
457
458 pct console 100
459
460Enter the LXC namespace and run a shell as root user
461
462 pct enter 100
463
464Display the configuration
465
466 pct config 100
467
468Add a network interface called eth0, bridged to the host bridge vmbr0,
469set the address and gateway, while it's running
470
471 pct set 100 -net0 name=eth0,bridge=vmbr0,ip=192.168.15.147/24,gw=192.168.15.1
472
473Reduce the memory of the container to 512MB
474
0585f29a
DM
475 pct set 100 -memory 512
476
04c569f6
DM
477
478Files
479------
480
481'/etc/pve/lxc/<CTID>.conf'::
482
483Configuration file for the container '<CTID>'.
484
485
0c6b782f
DM
486Container Advantages
487--------------------
488
489- Simple, and fully integrated into {pve}. Setup looks similar to a normal
490 VM setup.
491
492 * Storage (ZFS, LVM, NFS, Ceph, ...)
493
494 * Network
495
496 * Authentification
497
498 * Cluster
499
500- Fast: minimal overhead, as fast as bare metal
501
502- High density (perfect for idle workloads)
503
504- REST API
505
506- Direct hardware access
507
508
509Technology Overview
510-------------------
511
512- Integrated into {pve} graphical user interface (GUI)
513
514- LXC (https://linuxcontainers.org/)
515
516- cgmanager for cgroup management
517
518- lxcfs to provive containerized /proc file system
519
520- apparmor
521
522- CRIU: for live migration (planned)
523
11f340ff 524- We use latest available kernels (4.4.X)
0c6b782f 525
a8e99754 526- Image based deployment (templates)
0c6b782f
DM
527
528- Container setup from host (Network, DNS, Storage, ...)
529
530
531ifdef::manvolnum[]
532include::pve-copyright.adoc[]
533endif::manvolnum[]
534
535
536
537
538
539
540