Bug 1251943

What's your QEMU command line? You need to set werror=stop to -drive, that will make vm paused.

(In reply to Fam Zheng from comment #2)
> What's your QEMU command line? You need to set werror=stop to -drive, that
> will make vm paused.

when add "werror=stop" to -drive, guest can be stoped with "io-error", but before it be stoped, there still have flood iscsi error message as before.

command line:
/usr/libexec/qemu-kvm -name test -machine pc,accel=kvm,usb=off,dump-guest-core=on -m 6G -cpu SandyBridge -smp 4,sockets=4,cores=1,threads=1 -no-user-config -nodefaults -monitor stdio -rtc base=utc,driftfix=slew -global kvm-pit.lost_tick_policy=discard -no-hpet -global PIIX4_PM.disable_s3=1 -global PIIX4_PM.disable_s4=1 -boot menu=on,strict=on -device pci-bridge,bus=pci.0,id=bridge1,chassis_nr=1,addr=0x5 -device ich9-usb-ehci1,id=usb,bus=bridge1,addr=0x2.0x7 -device ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=bridge1,multifunction=on,addr=0x2.0x0 -device ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=bridge1,addr=0x2.0x1 -device ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=bridge1,addr=0x2.0x2 -device virtio-serial-pci,id=virtio-serial0,bus=bridge1,addr=0x4 -chardev socket,path=/tmp/yfu,server,nowait,id=yfu0 -netdev tap,id=hostnet,vhost=on -device virtio-net-pci,netdev=hostnet,id=net,mac=78:1a:4a:d6:b8:97,bus=bridge1,addr=0x6,bootindex=2 -vnc :1 -vga qxl -global qxl-vga.ram_size=67108864 -global qxl-vga.vram_size=33554432 -msg timestamp=on -monitor unix:/home/qmp,server,nowait -serial unix:/tmp/ttyS0,server,nowait -qmp tcp:0:4444,server,nowait -device virtio-scsi-pci,id=scsi -iscsi initiator-name=iqn.1994-05.com.redhat:c4836823271d -drive file=iscsi://10.66.4.122/fuyanan/2,if=none,id=drive-scsi,media=disk,format=qcow2,werror=stop,cache=none,aio=native,serial=sluo-disk1-iscsi,id=iqn -device scsi-disk,drive=drive-scsi,id=scsi-0,bus=scsi.0,bootindex=1 -drive file=/root/en_windows_8_1_enterprise_x64_dvd_2971902.iso,if=none,media=cdrom,readonly=on,format=raw,id=cdrom1 -device scsi-disk,drive=cdrom1,id=scsi-1,bus=scsi.0,bootindex=0 -fda /usr/share/virtio-win/virtio-win-1.7.4_amd64.vfd

What is the flood like?

(In reply to Fam Zheng from comment #4)
> What is the flood like?

sorry for make a mistake before.
1. without "werror=stop" ,i think the default value is "werror=enospc", qemu keep print:

2015-09-17T07:55:34.229860Z qemu-kvm: iSCSI Failure: SENSE KEY:(null)(3) ASCQ:(null)(0x1100)
2015-09-17T07:55:34.229986Z qemu-kvm: iSCSI Failure: SENSE KEY:(null)(3) ASCQ:(null)(0x1100)
2015-09-17T07:55:34.297813Z qemu-kvm: iSCSI Failure: SENSE KEY:ILLEGAL_REQUEST(5) ASCQ:LBA_OUT_OF_RANGE(0x2100)
2015-09-17T07:55:34.322869Z qemu-kvm: iSCSI Failure: SENSE KEY:ILLEGAL_REQUEST(5) ASCQ:LBA_OUT_OF_RANGE(0x2100)
2015-09-17T07:55:34.347555Z qemu-kvm: iSCSI Failure: SENSE KEY:ILLEGAL_REQUEST(5) ASCQ:LBA_OUT_OF_RANGE(0x2100)
......

and (qemu)info status,  to check the VM status is "runing"

2.with "werror=stop", qemu print a little then stop(these error messages can stop output after a little time):
2015-09-17T08:16:11.252753Z qemu-kvm: iSCSI Failure: SENSE KEY:(null)(3) ASCQ:(null)(0x1100)
2015-09-17T08:16:15.728537Z qemu-kvm: iSCSI Failure: SENSE KEY:(null)(3) ASCQ:(null)(0x1100)

then no output, after checked "info status", it is "VM status: paused (io-error)"

Issue confirmed.

When a write triggers cluster allocation, qcow2 uses the next unused cluster offset. It the offset falls beyond the end of iscsi target (right, this is not checked), we get the "LBA_OUT_OF_RANGE" error from iscsi target.

From iscsi driver's point of view, it is impossible to distinguish this particular case from a guest "out of range" access, as a result it's not right to return -ENOSPC here, instead -EIO is returned; on the other hand, qcow2 doesn't know whether the -EIO is because of "out of range" access or it is because other failures, due to the limitation of the interface.

I think iscsi driver should return -ERANGE for LBA_OUT_OF_RANGE error and qcow2 should translate -ERANGE to -ENOSPC.

This should go to upstream first.

Fam

This was fixed in upstream by commit:


commit e01dd3da5cf9aa90ae844d3b86c2c2762066edac
Author: Fam Zheng <famz>
Date:   Thu Nov 5 13:00:09 2015 +0800

    iscsi: Translate scsi sense into error code
    
    Previously we return -EIO blindly when anything goes wrong. Add a helper
    function to parse sense fields and try to make the return code more
    meaningful.
    
    This also fixes the default werror configuration (enospc) when we're
    using qcow2 on an iscsi lun. The old -EIO not being treated as out of
    space error failed to trigger vm stop.
    
    Signed-off-by: Fam Zheng <famz>
    Message-Id: <1446699609-11376-1-git-send-email-famz>
    [libiscsi 1.9 compatibility - Paolo]
    Signed-off-by: Paolo Bonzini <pbonzini>

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHBA-2016-2673.html