VirtualBox

Ticket #7127 (closed defect: fixed)

Opened 4 years ago

Last modified 3 years ago

iSCSI w/ QNAP targets gives read error / IDE controller reset / reboot of VM

Reported by: rschmied Owned by:
Priority: major Component: virtual disk
Version: VirtualBox 3.2.6 Keywords: iscsi storage
Cc: Guest type: Windows
Host type: Linux

Description

This is related to closed bug #6111 which has been closed. The initial problem (block i/o errors which has been seen on older versions of VirtualBox and the QNAP NAS s/w now shows more subtle but also severe problems:

  • XP SP3 fully patched running in VM
  • DMA on primary IDE channel for HDD is set to auto, does not go into 'DMA mode' but stays in 'PIO mode'
  • XP runs for a while, resets IDE controller several times
  • Machine reboots eventually

Here's a small excerpt from the log file, which is attached. The VM displays the 'The system has recovered from a serious error.' dialog box after XP has restarted.

01:49:19.035 PIIX3 ATA: execution time for ATA command 0xc4 was 9 seconds
01:49:19.035 PIIX3 ATA: Ctl#0: finished processing RESET
01:55:05.647 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:06.152 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:06.657 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:07.161 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:07.666 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:07.666 PIIX3 ATA: LUN#0: disk read error (rc=VERR_PARSE_ERROR iSector=0xf5cd38 cSectors=0x28)
01:55:08.675 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:42.383 PIIX3 ATA: Ctl#0: RESET, DevSel=0 AIOIf=0 CmdIf0=0xc5 (33694945 usec ago) CmdIf1=0x00 (-1 usec ago)
01:55:42.393 PIIX3 ATA: Async I/O thread probably stuck in operation, interrupting
01:55:42.393 PIIX3 ATA: LUN#0: disk write error (rc=VERR_TIMEOUT iSector=0x1285568 cSectors=0x40)
01:55:42.393 PIIX3 ATA: Ctl#0: finished processing RESET
01:55:42.393 PIIX3 ATA: execution time for ATA command 0xc5 was 33 seconds
01:55:42.398 iSCSI: login to target iqn.2004-04.com.qnap:ts-439:iscsi.xpsp3.8d49b8 successful
01:55:42.457 Changing the VM state from 'RUNNING' to 'RESETTING'.
01:55:42.547 CPUMSetGuestCpuIdFeature: Enabled APIC

QNAP Software is the latest available (3.3.0 0629T). The box reports the following information for the iSCSI daemon:

[~] # /sbin/iscsid --version
iscsid version 2.0-871
[~] # uname -a
Linux speicher 2.6.33.2 #1 SMP Tue Jun 29 06:46:33 CST 2010 i686 unknown
[~] # 

Attached the minidump of the Windows XP system and the relevant VBox.log file.

Attachments

DMP.zip Download (29.3 KB) - added by rschmied 4 years ago.
Windows XP mini dump
VBox.log Download (78.9 KB) - added by rschmied 4 years ago.
VBox log file
io-error-linux.png Download (36.6 KB) - added by rschmied 4 years ago.
I/O error on linux
xp-2010-07-07-20-16-50.log Download (49.9 KB) - added by rschmied 4 years ago.
VBox.log of the iSCSI disk running on 3.2.6 on a MacOS 10.6.4 MacBook Pro C2D

Change History

Changed 4 years ago by rschmied

Windows XP mini dump

Changed 4 years ago by rschmied

VBox log file

Changed 4 years ago by rschmied

I/O error on linux

comment:1 Changed 4 years ago by rschmied

I've checked with VirtualBox 3.2.6 on Mac OS 10.6.4 w/ VT-x enabled. I'm running Linux Karmic. It also shows I/O errors after a while. It seems as these are DMA related. See attached screen shot.

Changed 4 years ago by rschmied

VBox.log of the iSCSI disk running on 3.2.6 on a MacOS 10.6.4 MacBook Pro C2D

comment:2 Changed 4 years ago by rschmied

Attached log file is from an XP Machine on MBP that eventually froze running the same iSCSI disk as with the initial VBox.log running on Linux.

comment:3 Changed 4 years ago by rschmied

I'm wondering if anybody will have a look into this problem...? I was kind of excited when the initial problem #6111 went away for reasons that are beyond my knowledge (as I wrote either changes on the QNAP side or changes within VirtualBox). But the VM is still crashing eventually with a disk error and this renders the whole iSCSI subsystem useless for me. I really would appreciate a heads up here!

Thanks
-ralph

comment:4 Changed 4 years ago by aeichner

Is it possible to get a packet trace when things start to fail? Looking at the traffic might give some hints whats going wrong. Thank you.

comment:5 Changed 3 years ago by frank

  • Status changed from new to closed
  • Resolution set to fixed

No response, closing.

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use