Annotation of wikisrc/ports/xen/howto.mdwn, revision 1.155
1.144 maxv 1: [[!meta title="Xen HowTo"]]
2:
1.5 mspo 3: Introduction
1.13 gdt 4: ============
1.1 mspo 5:
1.149 maxv 6: Xen is a hypervisor for x86 hardware, which supports running multiple guest
1.58 gdt 7: operating systems on a single physical machine. Xen is a Type 1 or
8: bare-metal hypervisor; one uses the Xen kernel to control the CPU,
9: memory and console, a dom0 operating system which mediates access to
10: other hardware (e.g., disks, network, USB), and one or more domU
11: operating systems which operate in an unprivileged virtualized
12: environment. IO requests from the domU systems are forwarded by the
1.149 maxv 13: Xen hypervisor to the dom0 to be fulfilled.
1.12 gdt 14:
1.149 maxv 15: Xen supports different styles of guest:
16:
17: [[!table data="""
18: Style of guest |Supported by NetBSD
1.150 maxv 19: PV |Yes (dom0, domU)
20: HVM |Yes (domU)
1.149 maxv 21: PVHVM |No
22: PVH |No
23: """]]
24:
25: In Para-Virtualized (PV) mode, the guest OS does not attempt to access
26: hardware directly, but instead makes hypercalls to the hypervisor; PV
27: guests must be specifically coded for Xen. In HVM mode, no guest
28: modification is required; however, hardware support is required, such
29: as VT-x on Intel CPUs and SVM on AMD CPUs.
30:
1.27 jnemeth 31: At boot, the dom0 kernel is loaded as a module with Xen as the kernel.
1.12 gdt 32: The dom0 can start one or more domUs. (Booting is explained in detail
33: in the dom0 section.)
34:
35: This HOWTO presumes a basic familiarity with the Xen system
1.123 gdt 36: architecture, with installing NetBSD on i386/amd64 hardware, and with
37: installing software from pkgsrc. See also the [Xen
38: website](http://www.xenproject.org/).
1.1 mspo 39:
1.150 maxv 40: This HOWTO attempts to address both the case of running a NetBSD dom0
41: on hardware and running domUs under it (NetBSD and other), and also
42: running NetBSD as a domU in a VPS.
43:
1.15 gdt 44: Versions of Xen and NetBSD
45: ==========================
46:
1.27 jnemeth 47: Most of the installation concepts and instructions are independent
48: of Xen version and NetBSD version. This section gives advice on
49: which version to choose. Versions not in pkgsrc and older unsupported
50: versions of NetBSD are intentionally ignored.
1.15 gdt 51:
1.133 gdt 52: The term "amd64" is used to refer to both the NetBSD port and to the
1.150 maxv 53: hardware architecture on which it runs. Such hardware is generally
54: made by both Intel and AMD, and common on PC computers.
1.133 gdt 55:
1.146 maxv 56: Xen versions
57: ------------
1.15 gdt 58:
1.111 wiz 59: In NetBSD, Xen is provided in pkgsrc, via matching pairs of packages
1.15 gdt 60: xenkernel and xentools. We will refer only to the kernel versions,
61: but note that both packages must be installed together and must have
62: matching versions.
63:
1.145 maxv 64: Versions available in pkgsrc:
1.85 gdt 65:
1.145 maxv 66: [[!table data="""
67: Xen Version |Package Name |Xen CPU Support |EOL'ed By Upstream
68: 4.2 |xenkernel42 |32bit, 64bit |Yes
69: 4.5 |xenkernel45 |64bit |Yes
70: 4.6 |xenkernel46 |64bit |Partially
71: 4.8 |xenkernel48 |64bit |No
72: 4.11 |xenkernel411 |64bit |No
73: """]]
1.113 gdt 74:
1.96 gdt 75: See also the [Xen Security Advisory page](http://xenbits.xen.org/xsa/).
76:
1.145 maxv 77: Note: Xen 4.2 was the last version to support 32bit CPUs.
78:
1.147 maxv 79: NetBSD versions
80: ---------------
1.15 gdt 81:
1.147 maxv 82: NetBSD 8 is recommended as the stable version of the most recent
83: release for production use.
1.146 maxv 84:
85: For developing Xen, netbsd-current may be appropriate.
1.15 gdt 86:
87: As of NetBSD 6, a NetBSD domU will support multiple vcpus. There is
88: no SMP support for NetBSD as dom0. (The dom0 itself doesn't really
1.105 gdt 89: need SMP for dom0 functions; the lack of support is really a problem
90: when using a dom0 as a normal computer.)
1.15 gdt 91:
1.147 maxv 92: Note: NetBSD support is called XEN3. However, it does support Xen 4,
93: because the hypercall interface has remained identical.
94:
1.18 gdt 95: Architecture
96: ------------
97:
1.155 ! maxv 98: Xen itself runs on amd64 hardware. Practically, almost any computer
! 99: where one would want to run Xen today supports amd64.
! 100:
! 101: The dom0 system, plus each domU, can be either i386PAE or amd64.
! 102: i386 without PAE is not supported.
! 103:
! 104: The standard approach is to use NetBSD/amd64 for the dom0. For domUs,
! 105: NetBSD/i386 (PAE) and NetBSD/amd64 are in widespread use, and there is
1.135 gdt 106: little to no Xen-specific reason to prefer one over the other.
1.133 gdt 107:
108: Note that to use an i386 dom0 with Xen 4.5 or higher, one must build
1.135 gdt 109: (or obtain from pre-built packages) an amd64 Xen kernel and install
1.154 maxv 110: that on the system. Almost no one in the NetBSD/Xen community does
111: this, and the standard, well-tested, approach is to use an amd64 dom0.
1.135 gdt 112:
113: A [posting on
114: xen-devel](https://lists.xen.org/archives/html/xen-devel/2012-07/msg00085.html)
115: explained that PV system call overhead was higher on amd64, and thus
116: there is some notion that i386 guests are faster. It goes on to
117: caution that the total situation is complex and not entirely
118: understood. On top of that caution, the post is about Linux, not
1.155 ! maxv 119: NetBSD.
1.29 gdt 120:
1.15 gdt 121: NetBSD as a dom0
122: ================
123:
124: NetBSD can be used as a dom0 and works very well. The following
125: sections address installation, updating NetBSD, and updating Xen.
1.19 gdt 126: Note that it doesn't make sense to talk about installing a dom0 OS
127: without also installing Xen itself. We first address installing
128: NetBSD, which is not yet a dom0, and then adding Xen, pivoting the
129: NetBSD install to a dom0 install by just changing the kernel and boot
130: configuration.
1.15 gdt 131:
1.45 gdt 132: For experimenting with Xen, a machine with as little as 1G of RAM and
133: 100G of disk can work. For running many domUs in productions, far
1.135 gdt 134: more will be needed; e.g. 4-8G and 1T of disk is reasonable for a
135: half-dozen domUs of 512M and 32G each. Basically, the RAM and disk
136: have to be bigger than the sum of the RAM/disk needs of the dom0 and
137: all the domUs.
1.45 gdt 138:
1.142 gdt 139: In 2018-05, trouble booting a dom0 was reported with 256M of RAM: with
140: 512M it worked reliably. This does not make sense, but if you see
141: "not ELF" after Xen boots, try increasing dom0 RAM.
142:
1.15 gdt 143: Styles of dom0 operation
144: ------------------------
145:
146: There are two basic ways to use Xen. The traditional method is for
147: the dom0 to do absolutely nothing other than providing support to some
148: number of domUs. Such a system was probably installed for the sole
149: purpose of hosting domUs, and sits in a server room on a UPS.
150:
151: The other way is to put Xen under a normal-usage computer, so that the
152: dom0 is what the computer would have been without Xen, perhaps a
153: desktop or laptop. Then, one can run domUs at will. Purists will
154: deride this as less secure than the previous approach, and for a
155: computer whose purpose is to run domUs, they are right. But Xen and a
1.93 gdt 156: dom0 (without domUs) is not meaningfully less secure than the same
1.15 gdt 157: things running without Xen. One can boot Xen or boot regular NetBSD
158: alternately with little problems, simply refraining from starting the
159: Xen daemons when not running Xen.
160:
161: Note that NetBSD as dom0 does not support multiple CPUs. This will
1.51 gdt 162: limit the performance of the Xen/dom0 workstation approach. In theory
163: the only issue is that the "backend drivers" are not yet MPSAFE:
1.140 khorben 164: https://mail-index.netbsd.org/netbsd-users/2014/08/29/msg015195.html
1.15 gdt 165:
1.19 gdt 166: Installation of NetBSD
167: ----------------------
1.13 gdt 168:
1.19 gdt 169: First,
1.27 jnemeth 170: [install NetBSD/amd64](/guide/inst/)
1.19 gdt 171: just as you would if you were not using Xen.
172: However, the partitioning approach is very important.
173:
174: If you want to use RAIDframe for the dom0, there are no special issues
175: for Xen. Typically one provides RAID storage for the dom0, and the
1.22 gdt 176: domU systems are unaware of RAID. The 2nd-stage loader bootxx_* skips
1.111 wiz 177: over a RAID1 header to find /boot from a file system within a RAID
1.22 gdt 178: partition; this is no different when booting Xen.
1.19 gdt 179:
180: There are 4 styles of providing backing storage for the virtual disks
1.93 gdt 181: used by domUs: raw partitions, LVM, file-backed vnd(4), and SAN.
1.19 gdt 182:
183: With raw partitions, one has a disklabel (or gpt) partition sized for
184: each virtual disk to be used by the domU. (If you are able to predict
185: how domU usage will evolve, please add an explanation to the HOWTO.
186: Seriously, needs tend to change over time.)
187:
1.27 jnemeth 188: One can use [lvm(8)](/guide/lvm/) to create logical devices to use
189: for domU disks. This is almost as efficient as raw disk partitions
190: and more flexible. Hence raw disk partitions should typically not
191: be used.
1.19 gdt 192:
1.111 wiz 193: One can use files in the dom0 file system, typically created by dd'ing
1.19 gdt 194: /dev/zero to create a specific size. This is somewhat less efficient,
195: but very convenient, as one can cp the files for backup, or move them
196: between dom0 hosts.
197:
198: Finally, in theory one can place the files backing the domU disks in a
199: SAN. (This is an invitation for someone who has done this to add a
200: HOWTO page.)
1.1 mspo 201:
1.19 gdt 202: Installation of Xen
203: -------------------
1.1 mspo 204:
1.155 ! maxv 205: We will consider that you chose to use Xen 4.8, with NetBSD/amd64 as
! 206: dom0. In the dom0, install xenkernel48 and xentools48 from pkgsrc.
! 207: Ensure that your packages are recent.
! 208:
! 209: Once this is done, install the Xen kernel itself:
! 210:
! 211: [[!template id=programlisting text="""
! 212: # cp /usr/pkg/xen48-kernel/xen.gz /
! 213: """]]
! 214:
! 215: Then, place a NetBSD XEN3_DOM0 kernel in /, copied from
! 216: releasedir/amd64/binary/kernel/netbsd-XEN3_DOM0.gz of a NetBSD build.
1.20 gdt 217:
1.155 ! maxv 218: Add a line to /boot.cfg to boot Xen:
1.20 gdt 219:
1.152 maxv 220: [[!template id=programlisting text="""
221: menu=Xen:load /netbsd-XEN3_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
222: """]]
1.20 gdt 223:
1.155 ! maxv 224: which specifies that the dom0 should have 512MB of ram, leaving the rest
! 225: to be allocated for domUs. To use a serial console, use
1.77 gdt 226:
1.152 maxv 227: [[!template id=programlisting text="""
228: menu=Xen:load /netbsd-XEN3_DOM0.gz;multiboot /xen.gz dom0_mem=512M console=com1 com1=9600,8n1
229: """]]
1.77 gdt 230:
231: which will use the first serial port for Xen (which counts starting
1.139 gson 232: from 1, unlike NetBSD which counts starting from 0), forcing
233: speed/parity. Because the NetBSD command line lacks a
234: "console=pc" argument, it will use the default "xencons" console device,
235: which directs the console I/O through Xen to the same console device Xen
236: itself uses (in this case, the serial port).
237:
1.152 maxv 238: In an attempt to add performance, one can also add:
1.37 gdt 239:
1.152 maxv 240: [[!template id=programlisting text="""
241: dom0_max_vcpus=1 dom0_vcpus_pin
242: """]]
1.37 gdt 243:
244: to force only one vcpu to be provided (since NetBSD dom0 can't use
1.155 ! maxv 245: more) and to pin that vcpu to a physical CPU.
1.20 gdt 246:
1.93 gdt 247: Xen has [many boot
248: options](http://xenbits.xenproject.org/docs/4.5-testing/misc/xen-command-line.html),
1.111 wiz 249: and other than dom0 memory and max_vcpus, they are generally not
1.93 gdt 250: necessary.
251:
1.20 gdt 252: As with non-Xen systems, you should have a line to boot /netbsd (a
1.155 ! maxv 253: kernel that works without Xen).
1.127 gdt 254:
1.76 gdt 255: Now, reboot so that you are running a DOM0 kernel under Xen, rather
256: than GENERIC without Xen.
257:
1.54 gdt 258: Using grub (historic)
259: ---------------------
260:
261: Before NetBSD's native bootloader could support Xen, the use of
262: grub was recommended. If necessary, see the
1.135 gdt 263: [old grub information](/ports/xen/howto-grub).
1.54 gdt 264:
1.28 gdt 265: The [HowTo on Installing into
1.140 khorben 266: RAID-1](https://mail-index.NetBSD.org/port-xen/2006/03/01/0010.html)
1.28 gdt 267: explains how to set up booting a dom0 with Xen using grub with
268: NetBSD's RAIDframe. (This is obsolete with the use of NetBSD's native
1.135 gdt 269: boot. Now, just create a system with RAID-1, and alter /boot.cfg as
270: described above.)
1.28 gdt 271:
1.21 gdt 272: Configuring Xen
273: ---------------
274:
1.76 gdt 275: Now, you have a system that will boot Xen and the dom0 kernel, but not
276: do anything else special. Make sure that you have rebooted into Xen.
277: There will be no domUs, and none can be started because you still have
1.102 gdt 278: to configure the dom0 daemons.
1.21 gdt 279:
1.102 gdt 280: The daemons which should be run vary with Xen version and with whether
1.152 maxv 281: one is using xm or xl. Xen 4.2 and up packages use xl. To use xm with 4.2,
282: edit xendomains to use xm instead.
1.31 gdt 283:
1.132 gdt 284: For 4.1 and up, you should enable xencommons. Not enabling xencommons
285: will result in a hang; it is necessary to hit ^C on the console to let
286: the machine finish booting. If you are using xm (default in 4.1, or
287: if you changed xendomains in 4.2), you should also enable xend:
1.31 gdt 288:
1.152 maxv 289: [[!template id=programlisting text="""
290: xend=YES # only if using xm, and only installed <= 4.2
291: xencommons=YES
292: """]]
1.31 gdt 293:
294: TODO: Recommend for/against xen-watchdog.
1.27 jnemeth 295:
1.53 gdt 296: After you have configured the daemons and either started them (in the
1.79 gdt 297: order given) or rebooted, use xm or xl to inspect Xen's boot messages,
1.102 gdt 298: available resources, and running domains. An example with xl follows:
1.34 gdt 299:
1.153 maxv 300: [[!template id=programlisting text="""
301: # xl dmesg
302: ... xen's boot info ...
303: # xl info
304: ... available memory, etc ...
305: # xl list
306: Name Id Mem(MB) CPU State Time(s) Console
307: Domain-0 0 64 0 r---- 58.1
308: """]]
309:
310: Xen logs will be in /var/log/xen.
1.33 gdt 311:
1.88 gdt 312: ### Issues with xencommons
313:
314: xencommons starts xenstored, which stores data on behalf of dom0 and
315: domUs. It does not currently work to stop and start xenstored.
316: Certainly all domUs should be shutdown first, following the sort order
317: of the rc.d scripts. However, the dom0 sets up state with xenstored,
318: and is not notified when xenstored exits, leading to not recreating
319: the state when the new xenstored starts. Until there's a mechanism to
320: make this work, one should not expect to be able to restart xenstored
321: (and thus xencommons). There is currently no reason to expect that
322: this will get fixed any time soon.
323:
1.127 gdt 324: ### No-longer needed advice about devices
325:
326: The installation of NetBSD should already have created devices for xen
327: (xencons, xenevt, xsd_kva), but if they are not present, create them:
328:
1.153 maxv 329: [[!template id=programlisting text="""
330: cd /dev && sh MAKEDEV xen
331: """]]
1.127 gdt 332:
1.41 gdt 333: anita (for testing NetBSD)
334: --------------------------
335:
1.82 gdt 336: With the setup so far (assuming 4.2/xl), one should be able to run
337: anita (see pkgsrc/misc/py-anita) to test NetBSD releases, by doing (as
338: root, because anita must create a domU):
339:
1.153 maxv 340: [[!template id=programlisting text="""
341: anita --vmm=xl test file:///usr/obj/i386/
342: """]]
1.82 gdt 343:
344: Alternatively, one can use --vmm=xm to use xm-based domU creation
345: instead (and must, on Xen <= 4.1). TODO: confirm that anita xl really works.
346:
1.40 gdt 347: Xen-specific NetBSD issues
348: --------------------------
349:
350: There are (at least) two additional things different about NetBSD as a
351: dom0 kernel compared to hardware.
352:
1.111 wiz 353: One is that the module ABI is different because some of the #defines
1.109 gdt 354: change, so one must build modules for Xen. As of netbsd-7, the build
355: system does this automatically. TODO: check this. (Before building
356: Xen modules was added, it was awkward to use modules to the point
357: where it was considered that it did not work.)
1.40 gdt 358:
359: The other difference is that XEN3_DOM0 does not have exactly the same
360: options as GENERIC. While it is debatable whether or not this is a
361: bug, users should be aware of this and can simply add missing config
362: items if desired.
363:
1.15 gdt 364: Updating NetBSD in a dom0
365: -------------------------
366:
367: This is just like updating NetBSD on bare hardware, assuming the new
368: version supports the version of Xen you are running. Generally, one
369: replaces the kernel and reboots, and then overlays userland binaries
370: and adjusts /etc.
371:
372: Note that one must update both the non-Xen kernel typically used for
373: rescue purposes and the DOM0 kernel used with Xen.
374:
1.55 gdt 375: Converting from grub to /boot
376: -----------------------------
377:
378: These instructions were [TODO: will be] used to convert a system from
379: grub to /boot. The system was originally installed in February of
380: 2006 with a RAID1 setup and grub to boot Xen 2, and has been updated
381: over time. Before these commands, it was running NetBSD 6 i386, Xen
382: 4.1 and grub, much like the message linked earlier in the grub
383: section.
384:
1.111 wiz 385: # Install MBR bootblocks on both disks.
1.55 gdt 386: fdisk -i /dev/rwd0d
387: fdisk -i /dev/rwd1d
388: # Install NetBSD primary boot loader (/ is FFSv1) into RAID1 components.
389: installboot -v /dev/rwd0d /usr/mdec/bootxx_ffsv1
390: installboot -v /dev/rwd1d /usr/mdec/bootxx_ffsv1
391: # Install secondary boot loader
392: cp -p /usr/mdec/boot /
1.111 wiz 393: # Create boot.cfg following earlier guidance:
1.142 gdt 394: menu=Xen:load /netbsd-XEN3PAE_DOM0.gz console=pc;multiboot /xen.gz dom0_mem=512M
395: menu=Xen.ok:load /netbsd-XEN3PAE_DOM0.ok.gz console=pc;multiboot /xen.ok.gz dom0_mem=512M
1.55 gdt 396: menu=GENERIC:boot
397: menu=GENERIC single-user:boot -s
398: menu=GENERIC.ok:boot netbsd.ok
399: menu=GENERIC.ok single-user:boot netbsd.ok -s
400: menu=Drop to boot prompt:prompt
401: default=1
402: timeout=30
403:
404: TODO: actually do this and fix it if necessary.
1.22 gdt 405:
1.102 gdt 406: Upgrading Xen versions
1.15 gdt 407: ---------------------
408:
1.110 gdt 409: Minor version upgrades are trivial. Just rebuild/replace the
410: xenkernel version and copy the new xen.gz to / (where /boot.cfg
411: references it), and reboot.
412:
413: Major version upgrades are conceptually not difficult, but can run
414: into all the issues found when installing Xen. Assuming migration
415: from 4.1 to 4.2, remove the xenkernel41 and xentools41 packages and
416: install the xenkernel42 and xentools42 packages. Copy the 4.2 xen.gz
417: to /.
1.21 gdt 418:
1.102 gdt 419: Ensure that the contents of /etc/rc.d/xen* are correct. Specifically,
420: they must match the package you just installed and not be left over
421: from some previous installation.
422:
423: Enable the correct set of daemons; see the configuring section above.
424: (Upgrading from 3.x to 4.x without doing this will result in a hang.)
425:
426: Ensure that the domU config files are valid for the new version.
1.110 gdt 427: Specifically, for 4.x remove autorestart=True, and ensure that disks
428: are specified with numbers as the second argument, as the examples
429: above show, and not NetBSD device names.
1.15 gdt 430:
1.28 gdt 431: Unprivileged domains (domU)
432: ===========================
433:
434: This section describes general concepts about domUs. It does not
1.33 gdt 435: address specific domU operating systems or how to install them. The
436: config files for domUs are typically in /usr/pkg/etc/xen, and are
1.60 wiki 437: typically named so that the file name, domU name and the domU's host
1.33 gdt 438: name match.
439:
1.111 wiz 440: The domU is provided with CPU and memory by Xen, configured by the
1.33 gdt 441: dom0. The domU is provided with disk and network by the dom0,
442: mediated by Xen, and configured in the dom0.
443:
444: Entropy in domUs can be an issue; physical disks and network are on
445: the dom0. NetBSD's /dev/random system works, but is often challenged.
446:
1.48 gdt 447: Config files
448: ------------
449:
450: There is no good order to present config files and the concepts
451: surrounding what is being configured. We first show an example config
452: file, and then in the various sections give details.
453:
454: See (at least in xentools41) /usr/pkg/share/examples/xen/xmexample*,
455: for a large number of well-commented examples, mostly for running
456: GNU/Linux.
457:
458: The following is an example minimal domain configuration file
459: "/usr/pkg/etc/xen/foo". It is (with only a name change) an actual
460: known working config file on Xen 4.1 (NetBSD 5 amd64 dom0 and NetBSD 5
461: i386 domU). The domU serves as a network file server.
462:
463: # -*- mode: python; -*-
464:
465: kernel = "/netbsd-XEN3PAE_DOMU-i386-foo.gz"
466: memory = 1024
467: vif = [ 'mac=aa:00:00:d1:00:09,bridge=bridge0' ]
468: disk = [ 'file:/n0/xen/foo-wd0,0x0,w',
469: 'file:/n0/xen/foo-wd1,0x1,w' ]
470:
471: The domain will have the same name as the file. The kernel has the
472: host/domU name in it, so that on the dom0 one can update the various
473: domUs independently. The vif line causes an interface to be provided,
474: with a specific mac address (do not reuse MAC addresses!), in bridge
475: mode. Two disks are provided, and they are both writable; the bits
476: are stored in files and Xen attaches them to a vnd(4) device in the
1.111 wiz 477: dom0 on domain creation. The system treats xbd0 as the boot device
1.48 gdt 478: without needing explicit configuration.
479:
480: By default xm looks for domain config files in /usr/pkg/etc/xen. Note
481: that "xm create" takes the name of a config file, while other commands
482: take the name of a domain. To create the domain, connect to the
483: console, create the domain while attaching the console, shutdown the
484: domain, and see if it has finished stopping, do (or xl with Xen >=
485: 4.2):
486:
487: xm create foo
488: xm console foo
489: xm create -c foo
490: xm shutdown foo
1.90 gdt 491: xm list
1.48 gdt 492:
493: Typing ^] will exit the console session. Shutting down a domain is
494: equivalent to pushing the power button; a NetBSD domU will receive a
495: power-press event and do a clean shutdown. Shutting down the dom0
496: will trigger controlled shutdowns of all configured domUs.
497:
498: domU kernels
499: ------------
500:
501: On a physical computer, the BIOS reads sector 0, and a chain of boot
502: loaders finds and loads a kernel. Normally this comes from the root
1.111 wiz 503: file system. With Xen domUs, the process is totally different. The
1.48 gdt 504: normal path is for the domU kernel to be a file in the dom0's
1.111 wiz 505: file system. At the request of the dom0, Xen loads that kernel into a
1.48 gdt 506: new domU instance and starts execution. While domU kernels can be
507: anyplace, reasonable places to store domU kernels on the dom0 are in /
508: (so they are near the dom0 kernel), in /usr/pkg/etc/xen (near the
509: config files), or in /u0/xen (where the vdisks are).
510:
1.59 gdt 511: Note that loading the domU kernel from the dom0 implies that boot
512: blocks, /boot, /boot.cfg, and so on are all ignored in the domU.
1.48 gdt 513: See the VPS section near the end for discussion of alternate ways to
514: obtain domU kernels.
515:
1.33 gdt 516: CPU and memory
517: --------------
518:
1.48 gdt 519: A domain is provided with some number of vcpus, less than the number
1.111 wiz 520: of CPUs seen by the hypervisor. (For a dom0, this is controlled by
1.48 gdt 521: the boot argument "dom0_max_vcpus=1".) For a domU, it is controlled
522: from the config file by the "vcpus = N" directive.
523:
524: A domain is provided with memory; this is controlled in the config
525: file by "memory = N" (in megabytes). In the straightforward case, the
526: sum of the the memory allocated to the dom0 and all domUs must be less
1.33 gdt 527: than the available memory.
528:
529: Xen also provides a "balloon" driver, which can be used to let domains
530: use more memory temporarily. TODO: Explain better, and explain how
531: well it works with NetBSD.
1.28 gdt 532:
533: Virtual disks
534: -------------
535:
1.33 gdt 536: With the file/vnd style, typically one creates a directory,
537: e.g. /u0/xen, on a disk large enough to hold virtual disks for all
538: domUs. Then, for each domU disk, one writes zeros to a file that then
539: serves to hold the virtual disk's bits; a suggested name is foo-xbd0
540: for the first virtual disk for the domU called foo. Writing zeros to
541: the file serves two purposes. One is that preallocating the contents
542: improves performance. The other is that vnd on sparse files has
543: failed to work. TODO: give working/notworking NetBSD versions for
1.127 gdt 544: sparse vnd and gnats reference. Note that the use of file/vnd for Xen
545: is not really different than creating a file-backed virtual disk for
546: some other purpose, except that xentools handles the vnconfig
547: commands. To create an empty 4G virtual disk, simply do
1.39 gdt 548:
549: dd if=/dev/zero of=foo-xbd0 bs=1m count=4096
1.33 gdt 550:
1.89 gdt 551: Do not use qemu-img-xen, because this will create sparse file. There
552: have been recent (2015) reports of sparse vnd(4) devices causing
553: lockups, but there is apparently no PR.
554:
1.33 gdt 555: With the lvm style, one creates logical devices. They are then used
1.48 gdt 556: similarly to vnds. TODO: Add an example with lvm.
557:
558: In domU config files, the disks are defined as a sequence of 3-tuples.
559: The first element is "method:/path/to/disk". Common methods are
560: "file:" for file-backed vnd. and "phy:" for something that is already
561: a (TODO: character or block) device.
562:
563: The second element is an artifact of how virtual disks are passed to
564: Linux, and a source of confusion with NetBSD Xen usage. Linux domUs
565: are given a device name to associate with the disk, and values like
566: "hda1" or "sda1" are common. In a NetBSD domU, the first disk appears
567: as xbd0, the second as xbd1, and so on. However, xm/xl demand a
568: second argument. The name given is converted to a major/minor by
1.49 gdt 569: calling stat(2) on the name in /dev and this is passed to the domU.
570: In the general case, the dom0 and domU can be different operating
1.48 gdt 571: systems, and it is an unwarranted assumption that they have consistent
572: numbering in /dev, or even that the dom0 OS has a /dev. With NetBSD
573: as both dom0 and domU, using values of 0x0 for the first disk and 0x1
1.49 gdt 574: for the second works fine and avoids this issue. For a GNU/Linux
575: guest, one can create /dev/hda1 in /dev, or to pass 0x301 for
576: /dev/hda1.
1.48 gdt 577:
578: The third element is "w" for writable disks, and "r" for read-only
579: disks.
1.28 gdt 580:
1.127 gdt 581: Note that NetBSD by default creates only vnd[0123]. If you need more
582: than 4 total virtual disks at a time, run e.g. "./MAKEDEV vnd4" in the
583: dom0.
584:
585: Note that NetBSD by default creates only xbd[0123]. If you need more
586: virtual disks in a domU, run e.g. "./MAKEDEV xbd4" in the domU.
587:
1.28 gdt 588: Virtual Networking
589: ------------------
590:
1.111 wiz 591: Xen provides virtual Ethernets, each of which connects the dom0 and a
1.46 gdt 592: domU. For each virtual network, there is an interface "xvifN.M" in
593: the dom0, and in domU index N, a matching interface xennetM (NetBSD
594: name). The interfaces behave as if there is an Ethernet with two
1.111 wiz 595: adapters connected. From this primitive, one can construct various
1.46 gdt 596: configurations. We focus on two common and useful cases for which
597: there are existing scripts: bridging and NAT.
1.28 gdt 598:
1.48 gdt 599: With bridging (in the example above), the domU perceives itself to be
600: on the same network as the dom0. For server virtualization, this is
601: usually best. Bridging is accomplished by creating a bridge(4) device
602: and adding the dom0's physical interface and the various xvifN.0
603: interfaces to the bridge. One specifies "bridge=bridge0" in the domU
604: config file. The bridge must be set up already in the dom0; an
605: example /etc/ifconfig.bridge0 is:
1.46 gdt 606:
607: create
608: up
609: !brconfig bridge0 add wm0
1.28 gdt 610:
611: With NAT, the domU perceives itself to be behind a NAT running on the
612: dom0. This is often appropriate when running Xen on a workstation.
1.48 gdt 613: TODO: NAT appears to be configured by "vif = [ '' ]".
1.28 gdt 614:
1.49 gdt 615: The MAC address specified is the one used for the interface in the new
1.53 gdt 616: domain. The interface in dom0 will use this address XOR'd with
1.49 gdt 617: 00:00:00:01:00:00. Random MAC addresses are assigned if not given.
618:
1.33 gdt 619: Sizing domains
620: --------------
621:
622: Modern x86 hardware has vast amounts of resources. However, many
623: virtual servers can function just fine on far less. A system with
1.142 gdt 624: 512M of RAM and a 4G disk can be a reasonable choice. Note that it is
1.33 gdt 625: far easier to adjust virtual resources than physical ones. For
626: memory, it's just a config file edit and a reboot. For disk, one can
627: create a new file and vnconfig it (or lvm), and then dump/restore,
628: just like updating physical disks, but without having to be there and
629: without those pesky connectors.
630:
1.48 gdt 631: Starting domains automatically
632: ------------------------------
1.28 gdt 633:
1.48 gdt 634: To start domains foo at bar at boot and shut them down cleanly on dom0
635: shutdown, in rc.conf add:
1.28 gdt 636:
1.48 gdt 637: xendomains="foo bar"
1.28 gdt 638:
1.111 wiz 639: Note that earlier versions of the xentools41 xendomains rc.d script
640: used xl, when one should use xm with 4.1.
1.28 gdt 641:
642: Creating specific unprivileged domains (domU)
643: =============================================
1.14 gdt 644:
645: Creating domUs is almost entirely independent of operating system. We
1.49 gdt 646: have already presented the basics of config files. Note that you must
647: have already completed the dom0 setup so that "xl list" (or "xm list")
648: works.
1.14 gdt 649:
650: Creating an unprivileged NetBSD domain (domU)
651: ---------------------------------------------
1.1 mspo 652:
1.49 gdt 653: See the earlier config file, and adjust memory. Decide on how much
654: storage you will provide, and prepare it (file or lvm).
655:
1.111 wiz 656: While the kernel will be obtained from the dom0 file system, the same
1.49 gdt 657: file should be present in the domU as /netbsd so that tools like
658: savecore(8) can work. (This is helpful but not necessary.)
659:
660: The kernel must be specifically for Xen and for use as a domU. The
661: i386 and amd64 provide the following kernels:
662:
663: i386 XEN3PAE_DOMU
1.95 gdt 664: amd64 XEN3_DOMU
1.5 mspo 665:
1.49 gdt 666: This will boot NetBSD, but this is not that useful if the disk is
667: empty. One approach is to unpack sets onto the disk outside of xen
668: (by mounting it, just as you would prepare a physical disk for a
669: system you can't run the installer on).
670:
671: A second approach is to run an INSTALL kernel, which has a miniroot
672: and can load sets from the network. To do this, copy the INSTALL
673: kernel to / and change the kernel line in the config file to:
1.5 mspo 674:
1.49 gdt 675: kernel = "/home/bouyer/netbsd-INSTALL_XEN3_DOMU"
1.5 mspo 676:
1.49 gdt 677: Then, start the domain as "xl create -c configname".
1.1 mspo 678:
1.49 gdt 679: Alternatively, if you want to install NetBSD/Xen with a CDROM image, the following
680: line should be used in the config file.
1.1 mspo 681:
1.3 mspo 682: disk = [ 'phy:/dev/wd0e,0x1,w', 'phy:/dev/cd0a,0x2,r' ]
1.1 mspo 683:
684: After booting the domain, the option to install via CDROM may be
1.49 gdt 685: selected. The CDROM device should be changed to `xbd1d`.
1.1 mspo 686:
1.49 gdt 687: Once done installing, "halt -p" the new domain (don't reboot or halt,
688: it would reload the INSTALL_XEN3_DOMU kernel even if you changed the
689: config file), switch the config file back to the XEN3_DOMU kernel,
690: and start the new domain again. Now it should be able to use "root on
691: xbd0a" and you should have a, functional NetBSD domU.
1.1 mspo 692:
1.49 gdt 693: TODO: check if this is still accurate.
1.1 mspo 694: When the new domain is booting you'll see some warnings about *wscons*
695: and the pseudo-terminals. These can be fixed by editing the files
1.5 mspo 696: `/etc/ttys` and `/etc/wscons.conf`. You must disable all terminals in
697: `/etc/ttys`, except *console*, like this:
1.1 mspo 698:
1.3 mspo 699: console "/usr/libexec/getty Pc" vt100 on secure
700: ttyE0 "/usr/libexec/getty Pc" vt220 off secure
701: ttyE1 "/usr/libexec/getty Pc" vt220 off secure
702: ttyE2 "/usr/libexec/getty Pc" vt220 off secure
703: ttyE3 "/usr/libexec/getty Pc" vt220 off secure
1.1 mspo 704:
1.5 mspo 705: Finally, all screens must be commented out from `/etc/wscons.conf`.
1.1 mspo 706:
707: It is also desirable to add
708:
1.49 gdt 709: powerd=YES
1.1 mspo 710:
1.5 mspo 711: in rc.conf. This way, the domain will be properly shut down if
1.53 gdt 712: `xm shutdown -R` or `xm shutdown -H` is used on the dom0.
1.1 mspo 713:
1.92 gdt 714: It is not strictly necessary to have a kernel (as /netbsd) in the domU
1.111 wiz 715: file system. However, various programs (e.g. netstat) will use that
1.92 gdt 716: kernel to look up symbols to read from kernel virtual memory. If
717: /netbsd is not the running kernel, those lookups will fail. (This is
718: not really a Xen-specific issue, but because the domU kernel is
719: obtained from the dom0, it is far more likely to be out of sync or
720: missing with Xen.)
721:
1.14 gdt 722: Creating an unprivileged Linux domain (domU)
1.5 mspo 723: --------------------------------------------
1.1 mspo 724:
725: Creating unprivileged Linux domains isn't much different from
726: unprivileged NetBSD domains, but there are some details to know.
727:
728: First, the second parameter passed to the disk declaration (the '0x1' in
729: the example below)
730:
1.3 mspo 731: disk = [ 'phy:/dev/wd0e,0x1,w' ]
1.1 mspo 732:
733: does matter to Linux. It wants a Linux device number here (e.g. 0x300
1.49 gdt 734: for hda). Linux builds device numbers as: (major \<\< 8 + minor).
735: So, hda1 which has major 3 and minor 1 on a Linux system will have
736: device number 0x301. Alternatively, devices names can be used (hda,
737: hdb, ...) as xentools has a table to map these names to devices
738: numbers. To export a partition to a Linux guest we can use:
1.1 mspo 739:
1.49 gdt 740: disk = [ 'phy:/dev/wd0e,0x300,w' ]
741: root = "/dev/hda1 ro"
1.1 mspo 742:
743: and it will appear as /dev/hda on the Linux system, and be used as root
744: partition.
745:
1.49 gdt 746: To install the Linux system on the partition to be exported to the
747: guest domain, the following method can be used: install
748: sysutils/e2fsprogs from pkgsrc. Use mke2fs to format the partition
749: that will be the root partition of your Linux domain, and mount it.
750: Then copy the files from a working Linux system, make adjustments in
751: `/etc` (fstab, network config). It should also be possible to extract
752: binary packages such as .rpm or .deb directly to the mounted partition
753: using the appropriate tool, possibly running under NetBSD's Linux
1.111 wiz 754: emulation. Once the file system has been populated, umount it. If
755: desirable, the file system can be converted to ext3 using tune2fs -j.
1.49 gdt 756: It should now be possible to boot the Linux guest domain, using one of
757: the vmlinuz-\*-xenU kernels available in the Xen binary distribution.
1.1 mspo 758:
1.111 wiz 759: To get the Linux console right, you need to add:
1.1 mspo 760:
1.3 mspo 761: extra = "xencons=tty1"
1.1 mspo 762:
1.111 wiz 763: to your configuration since not all Linux distributions auto-attach a
1.1 mspo 764: tty to the xen console.
765:
1.14 gdt 766: Creating an unprivileged Solaris domain (domU)
1.5 mspo 767: ----------------------------------------------
1.1 mspo 768:
1.50 gdt 769: See possibly outdated
770: [Solaris domU instructions](/ports/xen/howto-solaris/).
1.5 mspo 771:
1.1 mspo 772:
1.52 gdt 773: PCI passthrough: Using PCI devices in guest domains
774: ---------------------------------------------------
1.1 mspo 775:
1.53 gdt 776: The dom0 can give other domains access to selected PCI
1.52 gdt 777: devices. This can allow, for example, a non-privileged domain to have
778: access to a physical network interface or disk controller. However,
779: keep in mind that giving a domain access to a PCI device most likely
780: will give the domain read/write access to the whole physical memory,
781: as PCs don't have an IOMMU to restrict memory access to DMA-capable
1.53 gdt 782: device. Also, it's not possible to export ISA devices to non-dom0
1.52 gdt 783: domains, which means that the primary VGA adapter can't be exported.
784: A guest domain trying to access the VGA registers will panic.
785:
1.53 gdt 786: If the dom0 is NetBSD, it has to be running Xen 3.1, as support has
1.52 gdt 787: not been ported to later versions at this time.
788:
789: For a PCI device to be exported to a domU, is has to be attached to
790: the "pciback" driver in dom0. Devices passed to the dom0 via the
791: pciback.hide boot parameter will attach to "pciback" instead of the
792: usual driver. The list of devices is specified as "(bus:dev.func)",
1.5 mspo 793: where bus and dev are 2-digit hexadecimal numbers, and func a
794: single-digit number:
1.1 mspo 795:
1.52 gdt 796: pciback.hide=(00:0a.0)(00:06.0)
1.1 mspo 797:
1.52 gdt 798: pciback devices should show up in the dom0's boot messages, and the
1.5 mspo 799: devices should be listed in the `/kern/xen/pci` directory.
1.1 mspo 800:
1.52 gdt 801: PCI devices to be exported to a domU are listed in the "pci" array of
802: the domU's config file, with the format "0000:bus:dev.func".
1.1 mspo 803:
1.52 gdt 804: pci = [ '0000:00:06.0', '0000:00:0a.0' ]
1.1 mspo 805:
1.52 gdt 806: In the domU an "xpci" device will show up, to which one or more pci
1.111 wiz 807: buses will attach. Then the PCI drivers will attach to PCI buses as
1.52 gdt 808: usual. Note that the default NetBSD DOMU kernels do not have "xpci"
809: or any PCI drivers built in by default; you have to build your own
810: kernel to use PCI devices in a domU. Here's a kernel config example;
811: note that only the "xpci" lines are unusual.
812:
813: include "arch/i386/conf/XEN3_DOMU"
814:
1.111 wiz 815: # Add support for PCI buses to the XEN3_DOMU kernel
1.52 gdt 816: xpci* at xenbus ?
817: pci* at xpci ?
818:
819: # PCI USB controllers
820: uhci* at pci? dev ? function ? # Universal Host Controller (Intel)
821:
822: # USB bus support
823: usb* at uhci?
824:
825: # USB Hubs
826: uhub* at usb?
827: uhub* at uhub? port ? configuration ? interface ?
828:
829: # USB Mass Storage
830: umass* at uhub? port ? configuration ? interface ?
831: wd* at umass?
832: # SCSI controllers
833: ahc* at pci? dev ? function ? # Adaptec [23]94x, aic78x0 SCSI
834:
835: # SCSI bus support (for both ahc and umass)
836: scsibus* at scsi?
837:
838: # SCSI devices
839: sd* at scsibus? target ? lun ? # SCSI disk drives
840: cd* at scsibus? target ? lun ? # SCSI CD-ROM drives
1.1 mspo 841:
842:
1.28 gdt 843: NetBSD as a domU in a VPS
844: =========================
845:
846: The bulk of the HOWTO is about using NetBSD as a dom0 on your own
847: hardware. This section explains how to deal with Xen in a domU as a
848: virtual private server where you do not control or have access to the
1.70 gdt 849: dom0. This is not intended to be an exhaustive list of VPS providers;
850: only a few are mentioned that specifically support NetBSD.
1.28 gdt 851:
1.52 gdt 852: VPS operators provide varying degrees of access and mechanisms for
853: configuration. The big issue is usually how one controls which kernel
1.111 wiz 854: is booted, because the kernel is nominally in the dom0 file system (to
855: which VPS users do not normally have access). A second issue is how
1.70 gdt 856: to install NetBSD.
1.52 gdt 857: A VPS user may want to compile a kernel for security updates, to run
858: npf, run IPsec, or any other reason why someone would want to change
859: their kernel.
860:
1.111 wiz 861: One approach is to have an administrative interface to upload a kernel,
1.68 gdt 862: or to select from a prepopulated list. Other approaches are pygrub
1.59 gdt 863: (deprecated) and pvgrub, which are ways to have a bootloader obtain a
1.111 wiz 864: kernel from the domU file system. This is closer to a regular physical
1.59 gdt 865: computer, where someone who controls a machine can replace the kernel.
1.52 gdt 866:
1.74 gdt 867: A second issue is multiple CPUs. With NetBSD 6, domUs support
868: multiple vcpus, and it is typical for VPS providers to enable multiple
869: CPUs for NetBSD domUs.
870:
1.68 gdt 871: pygrub
1.59 gdt 872: -------
1.52 gdt 873:
1.111 wiz 874: pygrub runs in the dom0 and looks into the domU file system. This
875: implies that the domU must have a kernel in a file system in a format
1.68 gdt 876: known to pygrub. As of 2014, pygrub seems to be of mostly historical
877: interest.
1.52 gdt 878:
1.59 gdt 879: pvgrub
880: ------
881:
882: pvgrub is a version of grub that uses PV operations instead of BIOS
883: calls. It is booted from the dom0 as the domU kernel, and then reads
1.111 wiz 884: /grub/menu.lst and loads a kernel from the domU file system.
1.59 gdt 885:
1.70 gdt 886: [Panix](http://www.panix.com/) lets users use pvgrub. Panix reports
1.71 gdt 887: that pvgrub works with FFsv2 with 16K/2K and 32K/4K block/frag sizes
888: (and hence with defaults from "newfs -O 2"). See [Panix's pvgrub
1.70 gdt 889: page](http://www.panix.com/v-colo/grub.html), which describes only
1.74 gdt 890: Linux but should be updated to cover NetBSD :-).
1.70 gdt 891:
892: [prgmr.com](http://prgmr.com/) also lets users with pvgrub to boot
893: their own kernel. See then [prgmr.com NetBSD
1.74 gdt 894: HOWTO](http://wiki.prgmr.com/mediawiki/index.php/NetBSD_as_a_DomU)
895: (which is in need of updating).
1.59 gdt 896:
1.70 gdt 897: It appears that [grub's FFS
898: code](http://xenbits.xensource.com/hg/xen-unstable.hg/file/bca284f67702/tools/libfsimage/ufs/fsys_ufs.c)
899: does not support all aspects of modern FFS, but there are also reports
1.72 gdt 900: that FFSv2 works fine. At prgmr, typically one has an ext2 or FAT
1.70 gdt 901: partition for the kernel with the intent that grub can understand it,
902: which leads to /netbsd not being the actual kernel. One must remember
1.111 wiz 903: to update the special boot partition.
1.59 gdt 904:
905: Amazon
906: ------
907:
1.143 wiki 908: See the [Amazon EC2 page](/amazon_ec2/).
1.44 gdt 909:
1.65 gdt 910: TODO items for improving NetBSD/xen
911: ===================================
912:
1.93 gdt 913: * Make the NetBSD dom0 kernel work with SMP.
914: * Test the Xen 4.5 packages adequately to be able to recommend them as
915: the standard approach.
916: * Get PCI passthrough working on Xen 4.5
1.65 gdt 917: * Get pvgrub into pkgsrc, either via xentools or separately.
918: * grub
1.70 gdt 919: * Check/add support to pkgsrc grub2 for UFS2 and arbitrary
1.66 gdt 920: fragsize/blocksize (UFS2 support may be present; the point is to
1.111 wiz 921: make it so that with any UFS1/UFS2 file system setup that works
1.66 gdt 922: with NetBSD grub will also work).
1.140 khorben 923: See [pkg/40258](https://gnats.netbsd.org/40258).
1.65 gdt 924: * Push patches upstream.
925: * Get UFS2 patches into pvgrub.
926: * Add support for PV ops to a version of /boot, and make it usable as
927: a kernel in Xen, similar to pvgrub.
1.93 gdt 928:
929: Random pointers
930: ===============
931:
1.117 gdt 932: This section contains links from elsewhere not yet integrated into the
933: HOWTO, and other guides.
1.93 gdt 934:
935: * http://www.lumbercartel.ca/library/xen/
936: * http://pbraun.nethence.com/doc/sysutils/xen_netbsd_dom0.html
1.117 gdt 937: * https://gmplib.org/~tege/xen.html
CVSweb for NetBSD wikisrc <wikimaster@NetBSD.org> software: FreeBSD-CVSweb