VirtualBox

Opened 13 years ago

Closed 4 years ago

#9466 closed defect (obsolete)

EXT4 warning: unaligned AIO/DIO

Reported by: Markus Duft Owned by:
Component: virtual disk Version: VirtualBox 4.1.2
Keywords: Cc:
Guest type: Windows Host type: Linux

Description (last modified by Klaus Espenlaub)

I just stumbled over this warning in /var/log/messages:

Aug 19 13:22:16 s01en22 kernel: [  144.131936] EXT4-fs (sdb1): Unaligned AIO/DIO on inode 7345581 by VBoxHeadless; performance will be poor.

any insights on what this could actually mean? i found two things about it on the web ([1] and [2]).

[1] http://www.spinics.net/lists/linux-ext4/msg22726.html
[2] http://ubuntuforums.org/showthread.php?t=1750417

Change History (12)

comment:1 by siegerstein, 12 years ago

same here... VB 4.1.6

[ 40.643570] EXT4-fs (sdb1): Unaligned AIO/DIO on inode 8126483 by VBoxHeadless; performance will be poor.

comment:2 by sergiomb, 12 years ago

I got this on VB 4.2.0: [22103.109265] warning: `VirtualBox' uses 32-bit capabilities (legacy support in use) [22127.185970] EXT4-fs (dm-1): Unaligned AIO/DIO on inode 530808 by AioMgr0-N; performance will be poor.

seems harmless, and I think is not new , but would like to know if we can improve performance somehow.

Last edited 12 years ago by sergiomb (previous) (diff)

comment:3 by jaapcrezee, 12 years ago

Nobody cares? I would like to know why VBox uses unaligned buffers and why or just fix it.

comment:4 by Klaus Espenlaub, 12 years ago

Description: modified (diff)

Of course we care, but several people seem to take those warnings far too seriously.

VirtualBox does not (and never did) any kind of re-buffering of read/write requests issued by the guest to guarantee alignment. This means that if the guest does unaligned reads/writes (and this can happens especially during initial bootstrap) such warnings are unavoidable.

It is only something to worry about if those warnings happen extremely frequently and during normal guest OS operation.

This can actually also happen generally with certain disk image formats (e.g. very old VDI images only used 512 byte alignment, more recent ones use 4K alignment and soon this will be increased again, and due to some allocation quirks VMDK can only achieve 2K alignment).

comment:5 by shishey, 11 years ago

I've just set up a Windows 2008 Server guest on an ubuntu 11.04 2.6.38-16-server host with VB 4.2.4. I'm encountering severe IO performance problems. An install step (writing in a database) that normally takes 15-20min took 5 hours. Similar setups with older version of VB but same hardware/OS/Guest-OS run fine.

The only difference is this warning:

EXT4-fs (dm-0): Unaligned AIO/DIO on inode XX by AioMgr0-N; performance will be poor.

I saw solutions like cloning the VDI, which does not seem to resolve the problem. Is there something I can do to fix the issue?

comment:6 by nix, 11 years ago

On Linux Guest too, kernel 3.2.44, 3.4.44, CentOS stock.

https://www.virtualbox.org/ticket/11797

comment:7 by sp, 11 years ago

Why can't a logged in user add themselves as cc: on a ticket?

comment:8 by Frank Mehnert, 11 years ago

Because the corresponding trac permission is not enabled for normal users. But this is not necessary. Adding a comment to a ticket is sufficient to get future notifications.

in reply to:  4 comment:9 by BBF, 10 years ago

Replying to klaus:

It is only something to worry about if those warnings happen extremely frequently and during normal guest OS operation.

No one will be able to know if these warnings are happening frequently or not since they are rate-limited to show up only once per day:

http://lxr.free-electrons.com/source/fs/ext4/file.c?v=3.3#L127

comment:10 by schneida, 10 years ago

I actually get quite a lot of those messages on my debian machine running a Windows 7 client, together with this (related??) trace:

[11797086.970088] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11818747.828175] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11840642.989830] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11862304.221048] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11883904.948856] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11905527.133646] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11924030.946557] composite sync not supported
[11924374.337111] device eth0 left promiscuous mode
[11924376.584196] vboxnetflt: dropped 676572 out of 21940020 packets
[11944644.500181] composite sync not supported
[11944663.453400] device eth0 entered promiscuous mode
[11944667.254121] EXT4-fs (dm-0): Unaligned AIO/DIO on inode 4064224 by VirtualBox; performance will be poor.
[11945701.368594] VirtualBox: page allocation failure: order:4, mode:0x20
[11945701.368598] Pid: 14859, comm: VirtualBox Tainted: G           O 3.2.0-4-486 #1 Debian 3.2.51-1
[11945701.368600] Call Trace:
[11945701.368606]  [<c107ae01>] ? warn_alloc_failed+0xc8/0xd9
[11945701.368609]  [<c107cb39>] ? __alloc_pages_nodemask+0x5bc/0x637
[11945701.368612]  [<c109da7c>] ? ____cache_alloc+0x214/0x3a6
[11945701.368635]  [<c109dc6d>] ? __kmalloc_track_caller+0x5f/0xa7
[11945701.368639]  [<c11e7f99>] ? alloc_skb+0x9/0xb
[11945701.368641]  [<c11e7eea>] ? __alloc_skb+0x50/0xf6
[11945701.368643]  [<c11e7f99>] ? alloc_skb+0x9/0xb
[11945701.368645]  [<c11e7fa8>] ? dev_alloc_skb+0xd/0x20
[11945701.368650]  [<f914dd87>] ? vboxNetFltLinuxSkBufFromSG+0x76/0x35f [vboxnetflt]
[11945701.368653]  [<f914e0d1>] ? vboxNetFltPortOsXmit+0x58/0x7e [vboxnetflt]
[11945701.368657]  [<f914ea39>] ? vboxNetFltPortXmit+0x129/0x147 [vboxnetflt]
[11945701.368673]  [<faa064ff>] ? VBoxHost_RTSpinlockReleaseNoInts+0xb/0xc [vboxdrv]
[11945701.368691]  [<fa9fddb4>] ? SUPR0ObjAddRefEx+0x166/0x179 [vboxdrv]
[11945701.368703]  [<faa0ad78>] ? VBoxHost_RTHandleTableLookupWithCtx+0x64/0x82 [vboxdrv]
[11945701.368709]  [<faa0ac3d>] ? rtHandleTableUnlock.isra.1+0xb/0xe [vboxdrv]
[11945701.368828]  [<faa011c8>] ? supdrvIOCtl+0x108e/0x1fe5 [vboxdrv]
[11945701.368836]  [<faa044eb>] ? rtR0MemAllocEx+0x4d/0x9c [vboxdrv]
[11945701.368841]  [<fa9fd18a>] ? SUPR0Printf+0x69/0x69 [vboxdrv]
[11945701.368846]  [<fa9ff8db>] ? supdrvIOCtlFast+0x59/0x61 [vboxdrv]
[11945701.368852]  [<fa9fd1b6>] ? VBoxDrvLinuxIOCtl_4_1_18+0x2c/0x172 [vboxdrv]
[11945701.368856]  [<c1039e18>] ? sched_clock_local.constprop.3+0xe/0x139
[11945701.368861]  [<fa9fd18a>] ? SUPR0Printf+0x69/0x69 [vboxdrv]
[11945701.368864]  [<c10af5df>] ? do_vfs_ioctl+0x45a/0x491
[11945701.368866]  [<c11370ae>] ? div_s64_rem+0x38/0x4c
[11945701.368869]  [<c10263df>] ? ns_to_timespec+0x1b/0x33
[11945701.368872]  [<c1037044>] ? sample_to_timespec+0x19/0x2a
[11945701.368875]  [<c128396f>] ? _cond_resched+0x5/0x18
[11945701.368877]  [<c10af65a>] ? sys_ioctl+0x44/0x66
[11945701.368880]  [<c1288463>] ? sysenter_do_call+0x12/0x28
[11945701.368882]  [<c1280000>] ? no_context+0x31/0x11c
[11945701.368884] Mem-Info:
[11945701.368885] DMA per-cpu:
[11945701.368887] CPU    0: hi:    0, btch:   1 usd:   0
[11945701.368888] Normal per-cpu:
[11945701.368889] CPU    0: hi:  186, btch:  31 usd:  63
[11945701.368891] HighMem per-cpu:
[11945701.368892] CPU    0: hi:  186, btch:  31 usd: 176
[11945701.368896] active_anon:348767 inactive_anon:114416 isolated_anon:0
[11945701.368897]  active_file:100841 inactive_file:90525 isolated_file:0
[11945701.368898]  unevictable:4 dirty:30 writeback:0 unstable:0
[11945701.368899]  free:123572 slab_reclaimable:14637 slab_unreclaimable:6079
[11945701.368900]  mapped:124040 shmem:2929 pagetables:2403 bounce:0
[11945701.368905] DMA free:3552kB min:64kB low:80kB high:96kB active_anon:0kB inactive_anon:0kB active_file:8184kB inactive_file:4084kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15804kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:68kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[11945701.368909] lowmem_reserve[]: 0 865 3539 3539
[11945701.368916] Normal free:292976kB min:3728kB low:4660kB high:5592kB active_anon:34416kB inactive_anon:105608kB active_file:120628kB inactive_file:89048kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:885944kB mlocked:0kB dirty:100kB writeback:0kB mapped:140552kB shmem:1552kB slab_reclaimable:58480kB slab_unreclaimable:24316kB kernel_stack:3600kB pagetables:428kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[11945701.368920] lowmem_reserve[]: 0 0 21391 21391
[11945701.368926] HighMem free:197760kB min:512kB low:3392kB high:6272kB active_anon:1360652kB inactive_anon:352056kB active_file:274552kB inactive_file:268968kB unevictable:16kB isolated(anon):0kB isolated(file):0kB present:2738064kB mlocked:16kB dirty:20kB writeback:0kB mapped:355608kB shmem:10164kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:9184kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[11945701.368931] lowmem_reserve[]: 0 0 0 0
[11945701.368933] DMA: 6*4kB 1*8kB 0*16kB 0*32kB 1*64kB 1*128kB 1*256kB 0*512kB 1*1024kB 1*2048kB 0*4096kB = 3552kB
[11945701.368939] Normal: 36634*4kB 12905*8kB 2596*16kB 52*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 292976kB
[11945701.368945] HighMem: 34634*4kB 6269*8kB 485*16kB 37*32kB 2*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 197760kB
[11945701.368950] 196729 total pagecache pages
[11945701.368952] 2434 pages in swap cache
[11945701.368953] Swap cache stats: add 217694, delete 215260, find 2730422/2740518
[11945701.368955] Free swap  = 8127728kB
[11945701.368956] Total swap = 8388604kB
[11945701.372004] 917216 pages RAM
[11945701.372004] 689906 pages HighMem
[11945701.372004] 109515 pages reserved
[11945701.372004] 259751 pages shared
[11945701.372004] 600702 pages non-shared

comment:11 by joyer, 8 years ago

Any updates?

Same problem here. Host: Debian7x64 Guest: Debian7x64

  1. Is there anything I have to do as the person who manages our server? For ex. to clone HDD images?
  2. How can I check if my vdi uses 512 byte alignment or not? "vboxmanage list hdds" shows no similart information.

comment:12 by aeichner, 4 years ago

Resolution: obsolete
Status: newclosed
Note: See TracTickets for help on using tickets.

© 2023 Oracle
ContactPrivacy policyTerms of Use