1virt-alignment-scan(1) Virtualization Support virt-alignment-scan(1)
2
6 virt-alignment-scan - Check alignment of virtual machine partitions
7
9 virt-alignment-scan [--options] -d domname
10 virt-alignment-scan [--options] -a disk.img [-a disk.img ...]
12 virt-alignment-scan [--options]
14
16 When older operating systems install themselves, the partitioning tools
17 place partitions at a sector misaligned with the underlying storage
18 (commonly the first partition starts on sector 63). Misaligned
19 partitions can result in an operating system issuing more I/O than
20 should be necessary.
21 The virt-alignment-scan tool checks the alignment of partitions in
23 virtual machines and disk images and warns you if there are alignment
24 problems.
25 Currently there is no virt tool for fixing alignment problems. You can
27 only reinstall the guest operating system. The following NetApp
28 document summarises the problem and possible solutions:
29 http://media.netapp.com/documents/tr-3747.pdf
30
32 To run this tool on a disk image directly, use the -a option:
33 $ virt-alignment-scan -a winxp.img
35 /dev/sda1 32256 512 bad (alignment < 4K)
36 $ virt-alignment-scan -a fedora16.img
38 /dev/sda1 1048576 1024K ok
39 /dev/sda2 2097152 2048K ok
40 /dev/sda3 526385152 2048K ok
41 To run the tool on a guest known to libvirt, use the -d option and
43 possibly the -c option:
44 # virt-alignment-scan -d RHEL5
46 /dev/sda1 32256 512 bad (alignment < 4K)
47 /dev/sda2 106928640 512 bad (alignment < 4K)
48 $ virt-alignment-scan -c qemu:///system -d Win7TwoDisks
50 /dev/sda1 1048576 1024K ok
51 /dev/sda2 105906176 1024K ok
52 /dev/sdb1 65536 64K ok
53 Run virt-alignment-scan without any -a or -d options to scan all
55 libvirt domains.
56 # virt-alignment-scan
58 F16x64:/dev/sda1 1048576 1024K ok
59 F16x64:/dev/sda2 2097152 2048K ok
60 F16x64:/dev/sda3 526385152 2048K ok
61 The output consists of 4 or more whitespace-separated columns. Only
63 the first 4 columns are significant if you want to parse this from a
64 program. The columns are:
65 col 1
67 The device and partition name (eg. /dev/sda1 meaning the first
68 partition on the first block device).
69 When listing all libvirt domains (no -a or -d option given) this
71 column is prefixed by the libvirt name or UUID (if --uuid is
72 given). eg: "WinXP:/dev/sda1"
73 col 2
75 the start of the partition in bytes
76 col 3
78 the alignment in bytes or Kbytes (eg. 512 or "4K")
79 col 4
81 "ok" if the alignment is best for performance, or "bad" if the
82 alignment can cause performance problems
83 cols 5+
85 optional free-text explanation.
86 The exit code from the program changes depending on whether poorly
88 aligned partitions were found. See "EXIT STATUS" below.
89 If you just want the exit code with no output, use the -q option.
91
93 --help
94 Display brief help.
95 -a file
97 --add file
98 Add file which should be a disk image from a virtual machine.
99 The format of the disk image is auto-detected. To override this
101 and force a particular format use the --format=.. option.
102 -a URI
104 --add URI
105 Add a remote disk. See "ADDING REMOTE STORAGE" in guestfish(1).
106 -c URI
108 --connect URI
109 If using libvirt, connect to the given URI. If omitted, then we
110 connect to the default libvirt hypervisor.
111 If you specify guest block devices directly (-a), then libvirt is
113 not used at all.
114 -d guest
116 --domain guest
117 Add all the disks from the named libvirt guest. Domain UUIDs can
118 be used instead of names.
119 --format=raw|qcow2|..
121 --format
122 The default for the -a option is to auto-detect the format of the
123 disk image. Using this forces the disk format for -a options which
124 follow on the command line. Using --format with no argument
125 switches back to auto-detection for subsequent -a options.
126 For example:
128 virt-alignment-scan --format=raw -a disk.img
130 forces raw format (no auto-detection) for disk.img.
132 virt-alignment-scan --format=raw -a disk.img --format -a another.img
134 forces raw format (no auto-detection) for disk.img and reverts to
136 auto-detection for another.img.
137 If you have untrusted raw-format guest disk images, you should use
139 this option to specify the disk format. This avoids a possible
140 security problem with malicious guests (CVE-2010-3851).
141 -P nr_threads
143 Since libguestfs 1.22, virt-alignment-scan is multithreaded and
144 examines guests in parallel. By default the number of threads to
145 use is chosen based on the amount of free memory available at the
146 time that virt-alignment-scan is started. You can force virt-
147 alignment-scan to use at most "nr_threads" by using the -P option.
148 Note that -P 0 means to autodetect, and -P 1 means to use a single
150 thread.
151 -q
153 --quiet
154 Don’t produce any output. Just set the exit code (see "EXIT
155 STATUS" below).
156 --uuid
158 Print UUIDs instead of names. This is useful for following a guest
159 even when the guest is migrated or renamed, or when two guests
160 happen to have the same name.
161 This option only applies when listing all libvirt domains (when no
163 -a or -d options are specified).
164 -v
166 --verbose
167 Enable verbose messages for debugging.
168 -V
170 --version
171 Display version number and exit.
172 -x Enable tracing of libguestfs API calls.
174
176 Operating systems older than Windows 2008 and Linux before ca.2010
177 place the first sector of the first partition at sector 63, with a 512
178 byte sector size. This happens because of a historical accident.
179 Drives have to report a cylinder / head / sector (CHS) geometry to the
180 BIOS. The geometry is completely meaningless on modern drives, but it
181 happens that the geometry reported always has 63 sectors per track.
182 The operating system therefore places the first partition at the start
183 of the second "track", at sector 63.
184 When the guest OS is virtualized, the host operating system and
186 hypervisor may prefer accesses aligned to one of:
187 · 512 bytes
189 if the host OS uses local storage directly on hard drive
191 partitions, and the hard drive has 512 byte physical sectors.
192 · 4 Kbytes
194 for local storage on new hard drives with 4Kbyte physical sectors;
196 for file-backed storage on filesystems with 4Kbyte block size; or
197 for some types of network-attached storage.
198 · 64 Kbytes
200 for high-end network-attached storage. This is the optimal block
202 size for some NetApp hardware.
203 · 1 Mbyte
205 see "1 MB PARTITION ALIGNMENT" below.
207 Partitions which are not aligned correctly to the underlying storage
209 cause extra I/O. For example:
210 sect#63
212 ┌──────────────────────────┬ ─ ─ ─ ─
213 │ guest │
214 │ filesystem block │
215 ─ ┬──────────────────┴──────┬───────────────────┴─────┬ ─ ─
216 │ host block │ host block │
217 │ │ │
218 ─ ┴─────────────────────────┴─────────────────────────┴ ─ ─
219 In this example, each time a 4K guest block is read, two blocks on the
221 host must be accessed (so twice as much I/O is done). When a 4K guest
222 block is written, two host blocks must first be read, the old and new
223 data combined, and the two blocks written back (4x I/O).
224 LINUX HOST BLOCK AND I/O SIZE
226 New versions of the Linux kernel expose the physical and logical block
227 size, and minimum and recommended I/O size.
228 For a typical consumer hard drive with 512 byte sectors:
230 $ cat /sys/block/sda/queue/hw_sector_size
232 512
233 $ cat /sys/block/sda/queue/physical_block_size
234 512
235 $ cat /sys/block/sda/queue/logical_block_size
236 512
237 $ cat /sys/block/sda/queue/minimum_io_size
238 512
239 $ cat /sys/block/sda/queue/optimal_io_size
240 0
241 For a new consumer hard drive with 4Kbyte sectors:
243 $ cat /sys/block/sda/queue/hw_sector_size
245 4096
246 $ cat /sys/block/sda/queue/physical_block_size
247 4096
248 $ cat /sys/block/sda/queue/logical_block_size
249 4096
250 $ cat /sys/block/sda/queue/minimum_io_size
251 4096
252 $ cat /sys/block/sda/queue/optimal_io_size
253 0
254 For a NetApp LUN:
256 $ cat /sys/block/sdc/queue/logical_block_size
258 512
259 $ cat /sys/block/sdc/queue/physical_block_size
260 512
261 $ cat /sys/block/sdc/queue/minimum_io_size
262 4096
263 $ cat /sys/block/sdc/queue/optimal_io_size
264 65536
265 The NetApp allows 512 byte accesses (but they will be very
267 inefficient), prefers a minimum 4K I/O size, but the optimal I/O size
268 is 64K.
269 For detailed information about what these numbers mean, see
271 http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/Storage_Administration_Guide/newstorage-iolimits.html
272 [Thanks to Matt Booth for providing 4K drive data. Thanks to Mike
274 Snitzer for providing NetApp data and additional information.]
275 1 MB PARTITION ALIGNMENT
277 Microsoft picked 1 MB as the default alignment for all partitions
278 starting with Windows 2008 Server, and Linux has followed this.
279 Assuming 512 byte sectors in the guest, you will now see the first
281 partition starting at sector 2048, and subsequent partitions (if any)
282 will start at a multiple of 2048 sectors.
283 1 MB alignment is compatible with all current alignment requirements
285 (4K, 64K) and provides room for future growth in physical block sizes.
286 SETTING ALIGNMENT
288 virt-resize(1) can change the alignment of the partitions of some
289 guests. Currently it can fully align all the partitions of all Windows
290 guests, and it will fix the bootloader where necessary. For Linux
291 guests, it can align the second and subsequent partitions, so the
292 majority of OS accesses except at boot will be aligned.
293 Another way to correct partition alignment problems is to reinstall
295 your guest operating systems. If you install operating systems from
296 templates, ensure these have correct partition alignment too.
297 For older versions of Windows, the following NetApp document contains
299 useful information: http://media.netapp.com/documents/tr-3747.pdf
300 For Red Hat Enterprise Linux ≤ 5, use a Kickstart script that contains
302 an explicit %pre section that creates aligned partitions using
303 parted(8). Do not use the Kickstart "part" command. The NetApp
304 document above contains an example.
305
307 This program returns:
308 · 0
310 successful exit, all partitions are aligned ≥ 64K for best
312 performance
313 · 1
315 an error scanning the disk image or guest
317 · 2
319 successful exit, some partitions have alignment < 64K which can
321 result in poor performance on high end network storage
322 · 3
324 successful exit, some partitions have alignment < 4K which can
326 result in poor performance on most hypervisors
327
329 guestfs(3), guestfish(1), virt-filesystems(1), virt-rescue(1),
330 virt-resize(1), http://libguestfs.org/.
331
333 Richard W.M. Jones http://people.redhat.com/~rjones/
334
336 Copyright (C) 2011 Red Hat Inc.
337
339 This program is free software; you can redistribute it and/or modify it
340 under the terms of the GNU General Public License as published by the
341 Free Software Foundation; either version 2 of the License, or (at your
342 option) any later version.
343 This program is distributed in the hope that it will be useful, but
345 WITHOUT ANY WARRANTY; without even the implied warranty of
346 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
347 General Public License for more details.
348 You should have received a copy of the GNU General Public License along
350 with this program; if not, write to the Free Software Foundation, Inc.,
351 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
352
354 To get a list of bugs against libguestfs, use this link:
355 https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools
356 To report a new bug against libguestfs, use this link:
358 https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools
359 When reporting a bug, please supply:
361 · The version of libguestfs.
363 · Where you got libguestfs (eg. which Linux distro, compiled from
365 source, etc)
366 · Describe the bug accurately and give a way to reproduce it.
368 · Run libguestfs-test-tool(1) and paste the complete, unedited output
370 into the bug report.
371libguestfs-1.40.1 2019-01-17 virt-alignment-scan(1)