« Return to Thread: IET chunk + ESXI5
Re: IET chunk + ESXI5
by Yucong Sun (叶雨飞)
::
Rate this Message:
// The first bad one
Target iqn.2010-09.com.outofwall:vstore3
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore3,ScsiSN=vstore3d0,IOMode=wt
// The good one
Target iqn.2010-09.com.outofwall:vstore2
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore2,ScsiSN=1,IOMode=wt
// The second bad one
Target iqn.2009-10.com.outofwall:vstore1
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore1,ScsiSN=vstore1D0,IOMode=wt
all reset of config are in default, as you see, scsisn problem is
fixed. I am not entirely sure they run on same version of ietd though,
I will try to load r470 on 1&3 to see whether it improve things.
The problem however will appear once in about 2 weeks, so it takes a
while to confirm that it is gone.
2012/4/2 Steffen Plotner <swplotner@...>:
> Hi,
>
> You are correct. Please also ensure that the scsi serial numbers are
> unique across ALL iscsi-target servers. Changing them will break
> connections on the esx side.
>
> Steffen
>
> -----Original Message-----
> From: "Yucong Sun (叶雨飞)" <sunyucong@...>
> Date: Mon, 2 Apr 2012 10:51:56 -0700
> To: Steffen Plotner <swplotner@...>
> Cc: "Ross S. W. Walker" <RWalker@...>, iscsitarget-devel
> <iscsitarget-devel@...>, Emmanuel Florac
> <eflorac@...>
> Subject: Re: [Iscsitarget-devel] IET chunk + ESXI5
>
>>let me try to explain again: I have 7 esxi server connecting to 3
>>iscsi server, and one of iscsi server are super solid, no problem
>>whatsoever, the other two have periodical problem with
>>
>>Mar 15 08:09:30 vstore-2 kernel: [656805.027625] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:32 vstore-2 kernel: [656807.039642] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:34 vstore-2 kernel: [656809.054532] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:35 vstore-2 kernel: [656809.879225] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:0 by sid:3662473236120064 (Function
>>Complete)
>>Mar 15 08:09:36 vstore-2 kernel: [656811.066953] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:38 vstore-2 kernel: [656813.083320] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>
>>Which then Ross points out it maybe caused by the fact that I don't
>>have an lun 0 on those two problematic servers, I fixed that by moving
>>the disk to lun 0, but the problem actually reappeared today. So I'm
>>thinking maybe it's caused by something else, like a counter wrap,
>>that why I propose to prevent deadlocking in this death mode, simply
>>reset the session after too many errors.
>>
>>
>>On Mon, Apr 2, 2012 at 10:47 AM, Steffen Plotner <swplotner@...>
>>wrote:
>>> Hi,
>>>
>>>> >> On Mar 17, 2012, at 7:48 PM, "Yucong Sun (叶雨飞)"
>>>> <sunyucong@...>
>>>> >> wrote:
>>>> >>
>>>> >> The problem is I have 2 machines that configured as such:
>>>> >>
>>>> >> Target iqn.2009-10.com.outofwall:vstore1
>>>> >> Lun 1
>>>> Path=/dev/sdb,Type=fileio,ScsiId=vstore1,ScsiSN=1,IOMode=wt
>>>> >>
>>>> >> Target iqn.2010-09.com.outofwall:vstore2
>>>> >> Lun 0
>>>> Path=/dev/sdb,Type=fileio,ScsiId=vstore2,ScsiSN=1,IOMode=wt
>>>> >>
>>>> >> and if I use the same Lun 0 , ESXi will think they are the
>>>> same device
>>>> >> address
>>>> >>
>>>
>>> I am not clear, if I am reading this correctly. You have 2 machines,
>>>what does that mean? You have two hosts accessing a single iscsi-target
>>>server or you have 2 iscsi-target servers? I am going assume you have a
>>>single iscsi-target server with 2 hosts attached, then it appears that
>>>you exposing /dev/sdb via 2 different targets which is generally not
>>>recommended, the underlying block device should only be opened once.
>>>
>>> Steffen
>
------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Iscsitarget-devel mailing list
Iscsitarget-devel@...
https://lists.sourceforge.net/lists/listinfo/iscsitarget-devel
Target iqn.2010-09.com.outofwall:vstore3
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore3,ScsiSN=vstore3d0,IOMode=wt
// The good one
Target iqn.2010-09.com.outofwall:vstore2
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore2,ScsiSN=1,IOMode=wt
// The second bad one
Target iqn.2009-10.com.outofwall:vstore1
Lun 0 Path=/dev/sdb,Type=fileio,ScsiId=vstore1,ScsiSN=vstore1D0,IOMode=wt
all reset of config are in default, as you see, scsisn problem is
fixed. I am not entirely sure they run on same version of ietd though,
I will try to load r470 on 1&3 to see whether it improve things.
The problem however will appear once in about 2 weeks, so it takes a
while to confirm that it is gone.
2012/4/2 Steffen Plotner <swplotner@...>:
> Hi,
>
> You are correct. Please also ensure that the scsi serial numbers are
> unique across ALL iscsi-target servers. Changing them will break
> connections on the esx side.
>
> Steffen
>
> -----Original Message-----
> From: "Yucong Sun (叶雨飞)" <sunyucong@...>
> Date: Mon, 2 Apr 2012 10:51:56 -0700
> To: Steffen Plotner <swplotner@...>
> Cc: "Ross S. W. Walker" <RWalker@...>, iscsitarget-devel
> <iscsitarget-devel@...>, Emmanuel Florac
> <eflorac@...>
> Subject: Re: [Iscsitarget-devel] IET chunk + ESXI5
>
>>let me try to explain again: I have 7 esxi server connecting to 3
>>iscsi server, and one of iscsi server are super solid, no problem
>>whatsoever, the other two have periodical problem with
>>
>>Mar 15 08:09:30 vstore-2 kernel: [656805.027625] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:32 vstore-2 kernel: [656807.039642] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:34 vstore-2 kernel: [656809.054532] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:35 vstore-2 kernel: [656809.879225] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:0 by sid:3662473236120064 (Function
>>Complete)
>>Mar 15 08:09:36 vstore-2 kernel: [656811.066953] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>Mar 15 08:09:38 vstore-2 kernel: [656813.083320] iscsi_trgt: Abort
>>Task (01) issued on tid:1 lun:1 by sid:3662473236120064 (Unknown LUN)
>>
>>Which then Ross points out it maybe caused by the fact that I don't
>>have an lun 0 on those two problematic servers, I fixed that by moving
>>the disk to lun 0, but the problem actually reappeared today. So I'm
>>thinking maybe it's caused by something else, like a counter wrap,
>>that why I propose to prevent deadlocking in this death mode, simply
>>reset the session after too many errors.
>>
>>
>>On Mon, Apr 2, 2012 at 10:47 AM, Steffen Plotner <swplotner@...>
>>wrote:
>>> Hi,
>>>
>>>> >> On Mar 17, 2012, at 7:48 PM, "Yucong Sun (叶雨飞)"
>>>> <sunyucong@...>
>>>> >> wrote:
>>>> >>
>>>> >> The problem is I have 2 machines that configured as such:
>>>> >>
>>>> >> Target iqn.2009-10.com.outofwall:vstore1
>>>> >> Lun 1
>>>> Path=/dev/sdb,Type=fileio,ScsiId=vstore1,ScsiSN=1,IOMode=wt
>>>> >>
>>>> >> Target iqn.2010-09.com.outofwall:vstore2
>>>> >> Lun 0
>>>> Path=/dev/sdb,Type=fileio,ScsiId=vstore2,ScsiSN=1,IOMode=wt
>>>> >>
>>>> >> and if I use the same Lun 0 , ESXi will think they are the
>>>> same device
>>>> >> address
>>>> >>
>>>
>>> I am not clear, if I am reading this correctly. You have 2 machines,
>>>what does that mean? You have two hosts accessing a single iscsi-target
>>>server or you have 2 iscsi-target servers? I am going assume you have a
>>>single iscsi-target server with 2 hosts attached, then it appears that
>>>you exposing /dev/sdb via 2 different targets which is generally not
>>>recommended, the underlying block device should only be opened once.
>>>
>>> Steffen
>
------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Iscsitarget-devel mailing list
Iscsitarget-devel@...
https://lists.sourceforge.net/lists/listinfo/iscsitarget-devel
« Return to Thread: IET chunk + ESXI5
| Free embeddable forum powered by Nabble | Forum Help |
