RE: failover does not work with rdac device handler

RE: failover does not work with rdac device handler

[Mario Mech]

Hi,

I need help with a similar or even the same issue as in the thread mentioned in the Subject 
(https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).

On two PowerEdge 2970 with CentOS 5.3 connected to a MD3000i I get the following messages in /var/log/messages

Jun 24 17:57:49 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:49 ninja kernel: end_request: I/O error, dev sdc, sector 530184
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:32.
Jun 24 17:57:49 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:49 ninja multipathd: 8:32: mark as failed
Jun 24 17:57:49 ninja multipathd: mpath8: remaining active paths: 3
Jun 24 17:57:49 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:49 ninja kernel: end_request: I/O error, dev sdd, sector 530184
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:48.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:80.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Could not failover device. Error 15.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:64.
Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Could not failover device. Error 15.
Jun 24 17:57:49 ninja kernel: GFS2: fsid=ninjo_cluster:gfsfs.0: gfs2_quotad: statfs error -5
Jun 24 17:57:49 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:50 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:50 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:50 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:50 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:50 ninja multipathd: 8:48: mark as failed
Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 2
Jun 24 17:57:50 ninja multipathd: 8:64: mark as failed
Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 1
Jun 24 17:57:50 ninja multipathd: 8:80: mark as failed
Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 0
Jun 24 17:57:54 ninja multipathd: 8:32: reinstated
Jun 24 17:57:54 ninja multipathd: mpath8: remaining active paths: 1
Jun 24 17:57:54 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:54 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:55 ninja multipathd: 8:64: reinstated
Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 2
Jun 24 17:57:55 ninja multipathd: 8:80: reinstated
Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 3
Jun 24 17:57:55 ninja multipathd: 8:48: reinstated
Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 4
Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered
Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered

The following additional information can I provide.

[root@ninja redhat]# lsmod | egrep 'rdac|multi'
scsi_dh_rdac           40897  1
dm_multipath           55257  2 dm_round_robin
scsi_dh                41665  2 scsi_dh_rdac,dm_multipath
dm_mod                100369  18 dm_multipath,dm_raid45,dm_snapshot,dm_zero,dm_mirror,dm_log
scsi_mod              196569  10 scsi_dh_rdac,ib_iser,iscsi_tcp,libiscsi,scsi_transport_iscsi,scsi_dh,sg,libata,megaraid_sas,sd_mod

[root@ninja redhat]# multipath -ll
mpath8 (360024e80004ec0de000010324a373eeb) dm-3 DELL,MD3000i
[size=7.5T][features=0][hwhandler=1 rdac][rw]
\_ round-robin 0 [prio=200][enabled]
  \_ 7:0:0:0  sde 8:64  [active][ghost]
  \_ 5:0:0:0  sdf 8:80  [active][ghost]
\_ round-robin 0 [prio=0][enabled]
  \_ 8:0:0:0  sdd 8:48  [active][ready]
  \_ 6:0:0:0  sdc 8:32  [active][ready]

The multipath version shipped with CentOS 5.3 is multipath-tools v0.4.7 (03/12, 2006).

As I've seen in the mentioned thread something like

echo rdac > /sys/block/sd?/device/dh_state

should fix at least the failover problem.

But I do not see any dh_state in /sys/block/sd?/device/ and I'm not allowed (even as root) to do anything like:

for i in c d e f; do echo rdac > /sys/block/sd$i/device/dh_state; done

At the moment I'd like to avoid building the multipath package by myself with the patch provided in the afore mentioned thread.

Do you have any ideas? Or is a different problem I have.

Every help is appreciated.

All best

Mario

-- 
Dr. Mario Mech

Institute for Geophysics and Meteorology
University of Cologne
Zuelpicherstr. 49a
50674 Cologne
Germany

t: +49 (0)221 - 470 - 1776
f: +49 (0)221 - 470 - 5198
e: mech@meteo.uni-koeln.de
w: http://www.meteo.uni-koeln.de/~mmech/

RE: failover does not work with rdac device handler

[Charlie Brady]

On Wed, 24 Jun 2009, Mario Mech wrote:

> I need help with a similar or even the same issue as in the thread mentioned 
> in the Subject 
> (https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).

I have the same issue occurring with a Sun StorageTek 2510 (iscsi 
storage array whose details and multipath.conf configuration I provided 
recently). AFAICS, the symptoms are identical.

It seems that it would be useful to be able to extend rdac_dev_list at 
module load time (e.g. via options in modprobe.conf). Thoughts?

---
Charlie

RE: failover does not work with rdac device handler

[Chandra Seetharaman]

On Wed, 2009-06-24 at 16:29 -0400, Charlie Brady wrote:
> On Wed, 24 Jun 2009, Mario Mech wrote:
> 
> > I need help with a similar or even the same issue as in the thread mentioned 
> > in the Subject 
> > (https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).
> 
> I have the same issue occurring with a Sun StorageTek 2510 (iscsi 
> storage array whose details and multipath.conf configuration I provided 
> recently). AFAICS, the symptoms are identical.
> 
> It seems that it would be useful to be able to extend rdac_dev_list at 
> module load time (e.g. via options in modprobe.conf). Thoughts?

That is a good idea. Let me think about it. 

OTOH, if you can roll up a patch, I will be happy to review :)
> 
> ---
> Charlie
> 
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel

RE: failover does not work with rdac device handler

[Chandra Seetharaman]

On Wed, 2009-06-24 at 18:09 +0200, Mario Mech wrote:
> Hi,
> 
> I need help with a similar or even the same issue as in the thread mentioned in the Subject 
> (https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).

My bad. I had some issues during debug and never got back to it.

Will send couple of patches by tomorrow to fix this problem.
 
> 
> On two PowerEdge 2970 with CentOS 5.3 connected to a MD3000i I get the following messages in /var/log/messages
> 
> Jun 24 17:57:49 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:49 ninja kernel: end_request: I/O error, dev sdc, sector 530184
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:32.
> Jun 24 17:57:49 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:49 ninja multipathd: 8:32: mark as failed
> Jun 24 17:57:49 ninja multipathd: mpath8: remaining active paths: 3
> Jun 24 17:57:49 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:49 ninja kernel: end_request: I/O error, dev sdd, sector 530184
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:48.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:80.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Could not failover device. Error 15.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Failing path 8:64.
> Jun 24 17:57:49 ninja kernel: device-mapper: multipath: Could not failover device. Error 15.
> Jun 24 17:57:49 ninja kernel: GFS2: fsid=ninjo_cluster:gfsfs.0: gfs2_quotad: statfs error -5
> Jun 24 17:57:49 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:50 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:50 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:50 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:50 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:50 ninja multipathd: 8:48: mark as failed
> Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 2
> Jun 24 17:57:50 ninja multipathd: 8:64: mark as failed
> Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 1
> Jun 24 17:57:50 ninja multipathd: 8:80: mark as failed
> Jun 24 17:57:50 ninja multipathd: mpath8: remaining active paths: 0
> Jun 24 17:57:54 ninja multipathd: 8:32: reinstated
> Jun 24 17:57:54 ninja multipathd: mpath8: remaining active paths: 1
> Jun 24 17:57:54 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:54 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:55 ninja multipathd: 8:64: reinstated
> Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 2
> Jun 24 17:57:55 ninja multipathd: 8:80: reinstated
> Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 3
> Jun 24 17:57:55 ninja multipathd: 8:48: reinstated
> Jun 24 17:57:55 ninja multipathd: mpath8: remaining active paths: 4
> Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered
> Jun 24 17:57:55 ninja multipathd: dm-3: add map (uevent)
> Jun 24 17:57:55 ninja multipathd: dm-3: devmap already registered
> 
> The following additional information can I provide.
> 
> [root@ninja redhat]# lsmod | egrep 'rdac|multi'
> scsi_dh_rdac           40897  1
> dm_multipath           55257  2 dm_round_robin
> scsi_dh                41665  2 scsi_dh_rdac,dm_multipath
> dm_mod                100369  18 dm_multipath,dm_raid45,dm_snapshot,dm_zero,dm_mirror,dm_log
> scsi_mod              196569  10 scsi_dh_rdac,ib_iser,iscsi_tcp,libiscsi,scsi_transport_iscsi,scsi_dh,sg,libata,megaraid_sas,sd_mod
> 
> [root@ninja redhat]# multipath -ll
> mpath8 (360024e80004ec0de000010324a373eeb) dm-3 DELL,MD3000i
> [size=7.5T][features=0][hwhandler=1 rdac][rw]
> \_ round-robin 0 [prio=200][enabled]
>   \_ 7:0:0:0  sde 8:64  [active][ghost]
>   \_ 5:0:0:0  sdf 8:80  [active][ghost]
> \_ round-robin 0 [prio=0][enabled]
>   \_ 8:0:0:0  sdd 8:48  [active][ready]
>   \_ 6:0:0:0  sdc 8:32  [active][ready]
> 
> The multipath version shipped with CentOS 5.3 is multipath-tools v0.4.7 (03/12, 2006).
> 
> As I've seen in the mentioned thread something like
> 
> echo rdac > /sys/block/sd?/device/dh_state
> 
> should fix at least the failover problem.
> 
> But I do not see any dh_state in /sys/block/sd?/device/ and I'm not allowed (even as root) to do anything like:
> 
> for i in c d e f; do echo rdac > /sys/block/sd$i/device/dh_state; done
> 
> At the moment I'd like to avoid building the multipath package by myself with the patch provided in the afore mentioned thread.

The changes are in kernel code.
> 
> Do you have any ideas? Or is a different problem I have.

It is the same problem and I am afraid that you do not have a way out.

May be next version of RHEL might add your device to rdac_dev_list (if
you could do that that would fix the problem).

> 
> Every help is appreciated.
> 
> All best
> 
> Mario
> 

Re: failover does not work with rdac device handler

[Mario Mech]

Chandra Seetharaman schrieb:
> On Wed, 2009-06-24 at 18:09 +0200, Mario Mech wrote:
>> Hi,
>>
>> I need help with a similar or even the same issue as in the thread mentioned in the Subject 
>> (https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).
> 
> My bad. I had some issues during debug and never got back to it.
> 
> Will send couple of patches by tomorrow to fix this problem.
>  

That sounds great.

>> At the moment I'd like to avoid building the multipath package by myself with the patch provided in the afore mentioned thread.
> 
> The changes are in kernel code.
>> Do you have any ideas? Or is a different problem I have.
> 
> It is the same problem and I am afraid that you do not have a way out.
> 
> May be next version of RHEL might add your device to rdac_dev_list (if
> you could do that that would fix the problem).
> 

I assume it's only building new kernel modules for my running 2.6.18-128.1.14.el5. That shouldn't be a big deal.

>> Every help is appreciated.
>>
>> All best
>>
>> Mario
>>
> 
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel


-- 
Dr. Mario Mech

Institute for Geophysics and Meteorology
University of Cologne
Zuelpicherstr. 49a
50674 Cologne
Germany

t: +49 (0)221 - 470 - 1776
f: +49 (0)221 - 470 - 5198
e: mech@meteo.uni-koeln.de
w: http://www.meteo.uni-koeln.de/~mmech/

Re: failover does not work with rdac device handler

[Chandra Seetharaman]

On Fri, 2009-06-26 at 11:24 +0200, Mario Mech wrote:
> Chandra Seetharaman schrieb:
> > On Wed, 2009-06-24 at 18:09 +0200, Mario Mech wrote:
> >> Hi,
> >>
> >> I need help with a similar or even the same issue as in the thread mentioned in the Subject 
> >> (https://www.redhat.com/archives/dm-devel/2008-October/msg00130.html).
> > 
> > My bad. I had some issues during debug and never got back to it.
> > 
> > Will send couple of patches by tomorrow to fix this problem.
> >  
> 
> That sounds great.
> 
> >> At the moment I'd like to avoid building the multipath package by myself with the patch provided in the afore mentioned thread.
> > 
> > The changes are in kernel code.
> >> Do you have any ideas? Or is a different problem I have.
> > 
> > It is the same problem and I am afraid that you do not have a way out.
> > 
> > May be next version of RHEL might add your device to rdac_dev_list (if
> > you could do that that would fix the problem).
> > 
> 
> I assume it's only building new kernel modules for my running 2.6.18-128.1.14.el5. That shouldn't be a big deal.

Yes. MD3000i is already in mainline code, you can just cut-n-paste that
entry for rdac_dev_list, compile the module and you are good to do.

chandra
> 
> >> Every help is appreciated.
> >>
> >> All best
> >>
> >> Mario
> >>
> > 
> > --
> > dm-devel mailing list
> > dm-devel@redhat.com
> > https://www.redhat.com/mailman/listinfo/dm-devel
> 
> 

Patch: add Dell product information into rdac device handler

[Yanqing_Liu]

[-- Attachment #1.1: Type: text/plain, Size: 515 bytes --]

Hi all, 

This patch is to add Dell Powervault storage arrays into device list for
rdac device handler.


--- linux-2.6.27-rc7/drivers/scsi/device_handler/scsi_dh_rdac.c.orig
2008-09-29 02:31:26.000000000 -0400
+++ linux-2.6.27-rc7/drivers/scsi/device_handler/scsi_dh_rdac.c
2008-09-30 02:06:26.000000000 -0400
@@ -590,6 +590,8 @@
 	{"STK", "OPENstorage D280"},
 	{"SUN", "CSM200_R"},
 	{"SUN", "LCSM100_F"},
+	{"DELL", "MD3000"},
+	{"DELL", "MD3000i"},
 	{NULL, NULL},
 };
 

Thanks,
Yanqing

[-- Attachment #1.2: Type: text/html, Size: 1923 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

failover does not work with rdac device handler

[Moger, Babu]

Hi All,

I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.

I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.

Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.

But I found that for some reason failover/failback is not working.  I am seeing following these messages.

"device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."

I can see that scsi_dh_rdac driver is already loaded.

I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).

My question - is it required to add all the supported vendor and model names in rdac_dev_list?

Appreciate your responses.

Thanks
Babu Moger

failover does not work with rdac device handler

[Moger, Babu]

Hi,

I have quick question.

I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).

Discovery seems fine. But, failover is not working.  I am seeing the following error.

"device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."

Is it required to add vendor id and model id anywhere else to support this device?

Thanks
Babu Moger


-----Original Message-----
From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
Sent: Wednesday, October 01, 2008 4:48 PM
To: device-mapper development; linux-scsi@vger.kernel.org
Subject: [dm-devel] failover does not work with rdac device handler

Hi All,

I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.

I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.

Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.

But I found that for some reason failover/failback is not working.  I am seeing following these messages.

"device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."

I can see that scsi_dh_rdac driver is already loaded.

I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).

My question - is it required to add all the supported vendor and model names in rdac_dev_list?

Appreciate your responses.

Thanks
Babu Moger







--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Re: [dm-devel] failover does not work with rdac device handler

[Chandra Seetharaman]

It is supposed to work. I will look into it and respond.

Can you try one thing 

for all your devices, do
	echo rdac > /sys/block/sd?/dh_state

(first you can read the files and see if they all show rdac or not)

and then do your fail over. Let me know what happens.

chandra
On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> Hi,
> 
> I have quick question.
> 
> I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
> 
> Discovery seems fine. But, failover is not working.  I am seeing the following error.
> 
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> 
> Is it required to add vendor id and model id anywhere else to support this device?
> 
> Thanks
> Babu Moger
> 
> 
> -----Original Message-----
> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> Sent: Wednesday, October 01, 2008 4:48 PM
> To: device-mapper development; linux-scsi@vger.kernel.org
> Subject: [dm-devel] failover does not work with rdac device handler
> 
> Hi All,
> 
> I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
> 
> I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
> 
> Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
> 
> But I found that for some reason failover/failback is not working.  I am seeing following these messages.
> 
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> 
> I can see that scsi_dh_rdac driver is already loaded.
> 
> I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
> 
> My question - is it required to add all the supported vendor and model names in rdac_dev_list?
> 
> Appreciate your responses.
> 
> Thanks
> Babu Moger
> 
> 
> 
> 
> 
> 
> 
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: failover does not work with rdac device handler

[Moger, Babu]

Thanks Chandra for the response..

That is the good news. So, we do not have to change the kernel  code everytime we want to support the new product.

I will do the steps you mentioned first thing on Monday and reply your back.

Thanks
Babu Moger

________________________________________
From: Chandra Seetharaman [sekharan@us.ibm.com]
Sent: Friday, October 03, 2008 7:50 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: Re: [dm-devel] failover does not work with rdac device handler

It is supposed to work. I will look into it and respond.

Can you try one thing

for all your devices, do
        echo rdac > /sys/block/sd?/dh_state

(first you can read the files and see if they all show rdac or not)

and then do your fail over. Let me know what happens.

chandra
On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> Hi,
>
> I have quick question.
>
> I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
>
> Discovery seems fine. But, failover is not working.  I am seeing the following error.
>
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
>
> Is it required to add vendor id and model id anywhere else to support this device?
>
> Thanks
> Babu Moger
>
>
> -----Original Message-----
> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> Sent: Wednesday, October 01, 2008 4:48 PM
> To: device-mapper development; linux-scsi@vger.kernel.org
> Subject: [dm-devel] failover does not work with rdac device handler
>
> Hi All,
>
> I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
>
> I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
>
> Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
>
> But I found that for some reason failover/failback is not working.  I am seeing following these messages.
>
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
>
> I can see that scsi_dh_rdac driver is already loaded.
>
> I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
>
> My question - is it required to add all the supported vendor and model names in rdac_dev_list?
>
> Appreciate your responses.
>
> Thanks
> Babu Moger
>
>
>
>
>
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: failover does not work with rdac device handler

[Moger, Babu]

Hi Chandra,

All the dh_state values show "rdac".  I have cut pasted the output below.

I am seeing the following error messages when I tried to failover.

        Oct  6 13:50:01 localhost multipathd: 8:96: reinstated
        Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Cannot              failover device because scsi_dh_rdac was not loaded.
        Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Failing             path 8:96.
        Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Could not           failover device. Error 15.
        Oct  6 13:50:01 localhost multipathd: 8:96: mark as failed


Output of # multipath -ll

        mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
        [size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
        \_ round-robin 0 [prio=6][active]
         \_ 3:0:0:2 sdf 8:80  [active][ready]
         \_ 3:0:2:2 sdl 8:176 [active][ready]
        \_ round-robin 0 [prio=0][enabled]
         \_ 3:0:1:2 sdi 8:128 [active][ghost]
         \_ 3:0:3:2 sdo 8:224 [active][ghost]


dh_state values.

        #cat /sys/block/sdf/device/dh_state
        rdac

        #cat /sys/block/sdl/device/dh_state
        rdac

        #cat /sys/block/sdi/device/dh_state
        rdac

        # cat /sys/block/sdo/device/dh_state
        rdac

 Please let me know what you think.
 Note (Kernel version 2.6.27-rc7 and multipath-tools v0.4.8)

Thanks
Babu Moger

-----Original Message-----
From: Moger, Babu
Sent: Friday, October 03, 2008 10:35 PM
To: sekharan@us.ibm.com
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler

Thanks Chandra for the response..

That is the good news. So, we do not have to change the kernel  code everytime we want to support the new product.

I will do the steps you mentioned first thing on Monday and reply your back.

Thanks
Babu Moger

________________________________________
From: Chandra Seetharaman [sekharan@us.ibm.com]
Sent: Friday, October 03, 2008 7:50 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: Re: [dm-devel] failover does not work with rdac device handler

It is supposed to work. I will look into it and respond.

Can you try one thing

for all your devices, do
        echo rdac > /sys/block/sd?/dh_state

(first you can read the files and see if they all show rdac or not)

and then do your fail over. Let me know what happens.

chandra
On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> Hi,
>
> I have quick question.
>
> I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
>
> Discovery seems fine. But, failover is not working.  I am seeing the following error.
>
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
>
> Is it required to add vendor id and model id anywhere else to support this device?
>
> Thanks
> Babu Moger
>
>
> -----Original Message-----
> From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> Sent: Wednesday, October 01, 2008 4:48 PM
> To: device-mapper development; linux-scsi@vger.kernel.org
> Subject: [dm-devel] failover does not work with rdac device handler
>
> Hi All,
>
> I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
>
> I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
>
> Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
>
> But I found that for some reason failover/failback is not working.  I am seeing following these messages.
>
> "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
>
> I can see that scsi_dh_rdac driver is already loaded.
>
> I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
>
> My question - is it required to add all the supported vendor and model names in rdac_dev_list?
>
> Appreciate your responses.
>
> Thanks
> Babu Moger
>
>
>
>
>
>
>
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [dm-devel] failover does not work with rdac device handler

[Chandra Seetharaman]

Is this the only rdac device that is present ?

Can you also get me the o/p of "lsmod | grep scsi_dh_rdac" (just wanted
to make sure if each (SCSI) device has a module.

On Mon, 2008-10-06 at 13:26 -0600, Moger, Babu wrote:
> Hi Chandra,
> 
> All the dh_state values show "rdac".  I have cut pasted the output below.
> 
> I am seeing the following error messages when I tried to failover.
> 
>         Oct  6 13:50:01 localhost multipathd: 8:96: reinstated
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Cannot              failover device because scsi_dh_rdac was not loaded.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Failing             path 8:96.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Could not           failover device. Error 15.
>         Oct  6 13:50:01 localhost multipathd: 8:96: mark as failed
> 
> 
> Output of # multipath -ll
> 
>         mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
>         [size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
>         \_ round-robin 0 [prio=6][active]
>          \_ 3:0:0:2 sdf 8:80  [active][ready]
>          \_ 3:0:2:2 sdl 8:176 [active][ready]
>         \_ round-robin 0 [prio=0][enabled]
>          \_ 3:0:1:2 sdi 8:128 [active][ghost]
>          \_ 3:0:3:2 sdo 8:224 [active][ghost]
> 
> 
> dh_state values.
> 
>         #cat /sys/block/sdf/device/dh_state
>         rdac
> 
>         #cat /sys/block/sdl/device/dh_state
>         rdac
> 
>         #cat /sys/block/sdi/device/dh_state
>         rdac
> 
>         # cat /sys/block/sdo/device/dh_state
>         rdac
> 
>  Please let me know what you think.
>  Note (Kernel version 2.6.27-rc7 and multipath-tools v0.4.8)
> 
> Thanks
> Babu Moger
> 
> -----Original Message-----
> From: Moger, Babu
> Sent: Friday, October 03, 2008 10:35 PM
> To: sekharan@us.ibm.com
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: RE: [dm-devel] failover does not work with rdac device handler
> 
> Thanks Chandra for the response..
> 
> That is the good news. So, we do not have to change the kernel  code everytime we want to support the new product.
> 
> I will do the steps you mentioned first thing on Monday and reply your back.
> 
> Thanks
> Babu Moger
> 
> ________________________________________
> From: Chandra Seetharaman [sekharan@us.ibm.com]
> Sent: Friday, October 03, 2008 7:50 PM
> To: Moger, Babu
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: Re: [dm-devel] failover does not work with rdac device handler
> 
> It is supposed to work. I will look into it and respond.
> 
> Can you try one thing
> 
> for all your devices, do
>         echo rdac > /sys/block/sd?/dh_state
> 
> (first you can read the files and see if they all show rdac or not)
> 
> and then do your fail over. Let me know what happens.
> 
> chandra
> On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> > Hi,
> >
> > I have quick question.
> >
> > I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
> >
> > Discovery seems fine. But, failover is not working.  I am seeing the following error.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > Is it required to add vendor id and model id anywhere else to support this device?
> >
> > Thanks
> > Babu Moger
> >
> >
> > -----Original Message-----
> > From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> > Sent: Wednesday, October 01, 2008 4:48 PM
> > To: device-mapper development; linux-scsi@vger.kernel.org
> > Subject: [dm-devel] failover does not work with rdac device handler
> >
> > Hi All,
> >
> > I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
> >
> > I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
> >
> > Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
> >
> > But I found that for some reason failover/failback is not working.  I am seeing following these messages.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > I can see that scsi_dh_rdac driver is already loaded.
> >
> > I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
> >
> > My question - is it required to add all the supported vendor and model names in rdac_dev_list?
> >
> > Appreciate your responses.
> >
> > Thanks
> > Babu Moger
> >
> >
> >
> >
> >
> >
> >
> > --
> > dm-devel mailing list
> > dm-devel@redhat.com
> > https://www.redhat.com/mailman/listinfo/dm-devel
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: failover does not work with rdac device handler

[Moger, Babu]

I have configured two Luns.


# lsmod | grep scsi_dh_rdac
scsi_dh_rdac           39044  8
scsi_dh                40068  2 scsi_dh_rdac,dm_multipath
scsi_mod              188920  9 scsi_dh_rdac,scsi_dh,sg,lpfc,qla2xxx,scsi_transport_fc,libata,megaraid_sas,sd_mod


Here is the complete output.

# multipath -ll

mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
[size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
\_ round-robin 0 [prio=6][active]
 \_ 3:0:0:2 sdf 8:80  [active][ready]
 \_ 3:0:2:2 sdl 8:176 [active][ready]
\_ round-robin 0 [prio=0][enabled]
 \_ 3:0:1:2 sdi 8:128 [active][ghost]
 \_ 3:0:3:2 sdo 8:224 [active][ghost]

mpath3 (3600a0b80000f519c00007e034714d159) dm-3 LSI,INF-01-00
[size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
\_ round-robin 0 [prio=6][active]
 \_ 3:0:1:3 sdj 8:144 [active][ready]
 \_ 3:0:3:3 sdp 8:240 [active][ready]
\_ round-robin 0 [prio=0][enabled]
 \_ 3:0:0:3 sdg 8:96  [active][ghost]
 \_ 3:0:2:3 sdm 8:192 [active][ghost]


Thanks
Babu Moger

-----Original Message-----
From: Chandra Seetharaman [mailto:sekharan@us.ibm.com]
Sent: Monday, October 06, 2008 2:54 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler

Is this the only rdac device that is present ?

Can you also get me the o/p of "lsmod | grep scsi_dh_rdac" (just wanted
to make sure if each (SCSI) device has a module.

On Mon, 2008-10-06 at 13:26 -0600, Moger, Babu wrote:
> Hi Chandra,
>
> All the dh_state values show "rdac".  I have cut pasted the output below.
>
> I am seeing the following error messages when I tried to failover.
>
>         Oct  6 13:50:01 localhost multipathd: 8:96: reinstated
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Cannot              failover device because scsi_dh_rdac was not loaded.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Failing             path 8:96.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Could not           failover device. Error 15.
>         Oct  6 13:50:01 localhost multipathd: 8:96: mark as failed
>
>
> Output of # multipath -ll
>
>         mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
>         [size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
>         \_ round-robin 0 [prio=6][active]
>          \_ 3:0:0:2 sdf 8:80  [active][ready]
>          \_ 3:0:2:2 sdl 8:176 [active][ready]
>         \_ round-robin 0 [prio=0][enabled]
>          \_ 3:0:1:2 sdi 8:128 [active][ghost]
>          \_ 3:0:3:2 sdo 8:224 [active][ghost]
>
>
> dh_state values.
>
>         #cat /sys/block/sdf/device/dh_state
>         rdac
>
>         #cat /sys/block/sdl/device/dh_state
>         rdac
>
>         #cat /sys/block/sdi/device/dh_state
>         rdac
>
>         # cat /sys/block/sdo/device/dh_state
>         rdac
>
>  Please let me know what you think.
>  Note (Kernel version 2.6.27-rc7 and multipath-tools v0.4.8)
>
> Thanks
> Babu Moger
>
> -----Original Message-----
> From: Moger, Babu
> Sent: Friday, October 03, 2008 10:35 PM
> To: sekharan@us.ibm.com
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: RE: [dm-devel] failover does not work with rdac device handler
>
> Thanks Chandra for the response..
>
> That is the good news. So, we do not have to change the kernel  code everytime we want to support the new product.
>
> I will do the steps you mentioned first thing on Monday and reply your back.
>
> Thanks
> Babu Moger
>
> ________________________________________
> From: Chandra Seetharaman [sekharan@us.ibm.com]
> Sent: Friday, October 03, 2008 7:50 PM
> To: Moger, Babu
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: Re: [dm-devel] failover does not work with rdac device handler
>
> It is supposed to work. I will look into it and respond.
>
> Can you try one thing
>
> for all your devices, do
>         echo rdac > /sys/block/sd?/dh_state
>
> (first you can read the files and see if they all show rdac or not)
>
> and then do your fail over. Let me know what happens.
>
> chandra
> On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> > Hi,
> >
> > I have quick question.
> >
> > I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
> >
> > Discovery seems fine. But, failover is not working.  I am seeing the following error.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > Is it required to add vendor id and model id anywhere else to support this device?
> >
> > Thanks
> > Babu Moger
> >
> >
> > -----Original Message-----
> > From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> > Sent: Wednesday, October 01, 2008 4:48 PM
> > To: device-mapper development; linux-scsi@vger.kernel.org
> > Subject: [dm-devel] failover does not work with rdac device handler
> >
> > Hi All,
> >
> > I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
> >
> > I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
> >
> > Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
> >
> > But I found that for some reason failover/failback is not working.  I am seeing following these messages.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > I can see that scsi_dh_rdac driver is already loaded.
> >
> > I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
> >
> > My question - is it required to add all the supported vendor and model names in rdac_dev_list?
> >
> > Appreciate your responses.
> >
> > Thanks
> > Babu Moger
> >
> >
> >
> >
> >
> >
> >
> > --
> > dm-devel mailing list
> > dm-devel@redhat.com
> > https://www.redhat.com/mailman/listinfo/dm-devel
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: failover does not work with rdac device handler

[Moger, Babu]

Hi Chandra,

Please note that failover works fine if I add vendor and model ids("LSI", "INF-01-00")in rdac_dev_list(in the file scsi_dh_rdac.c).

Thanks
Babu Moger

-----Original Message-----
From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
Sent: Monday, October 06, 2008 3:00 PM
To: sekharan@us.ibm.com
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler


I have configured two Luns.


# lsmod | grep scsi_dh_rdac
scsi_dh_rdac           39044  8
scsi_dh                40068  2 scsi_dh_rdac,dm_multipath
scsi_mod              188920  9 scsi_dh_rdac,scsi_dh,sg,lpfc,qla2xxx,scsi_transport_fc,libata,megaraid_sas,sd_mod


Here is the complete output.

# multipath -ll

mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
[size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
\_ round-robin 0 [prio=6][active]
 \_ 3:0:0:2 sdf 8:80  [active][ready]
 \_ 3:0:2:2 sdl 8:176 [active][ready]
\_ round-robin 0 [prio=0][enabled]
 \_ 3:0:1:2 sdi 8:128 [active][ghost]
 \_ 3:0:3:2 sdo 8:224 [active][ghost]

mpath3 (3600a0b80000f519c00007e034714d159) dm-3 LSI,INF-01-00
[size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
\_ round-robin 0 [prio=6][active]
 \_ 3:0:1:3 sdj 8:144 [active][ready]
 \_ 3:0:3:3 sdp 8:240 [active][ready]
\_ round-robin 0 [prio=0][enabled]
 \_ 3:0:0:3 sdg 8:96  [active][ghost]
 \_ 3:0:2:3 sdm 8:192 [active][ghost]


Thanks
Babu Moger

-----Original Message-----
From: Chandra Seetharaman [mailto:sekharan@us.ibm.com]
Sent: Monday, October 06, 2008 2:54 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler

Is this the only rdac device that is present ?

Can you also get me the o/p of "lsmod | grep scsi_dh_rdac" (just wanted
to make sure if each (SCSI) device has a module.

On Mon, 2008-10-06 at 13:26 -0600, Moger, Babu wrote:
> Hi Chandra,
>
> All the dh_state values show "rdac".  I have cut pasted the output below.
>
> I am seeing the following error messages when I tried to failover.
>
>         Oct  6 13:50:01 localhost multipathd: 8:96: reinstated
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Cannot              failover device because scsi_dh_rdac was not loaded.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Failing             path 8:96.
>         Oct  6 13:50:01 localhost kernel: device-mapper: multipath: Could not           failover device. Error 15.
>         Oct  6 13:50:01 localhost multipathd: 8:96: mark as failed
>
>
> Output of # multipath -ll
>
>         mpath2 (3600a0b80000f6a7d000080454714d058) dm-2 LSI,INF-01-00
>         [size=10G][features=1 queue_if_no_path][hwhandler=1 rdac][rw]
>         \_ round-robin 0 [prio=6][active]
>          \_ 3:0:0:2 sdf 8:80  [active][ready]
>          \_ 3:0:2:2 sdl 8:176 [active][ready]
>         \_ round-robin 0 [prio=0][enabled]
>          \_ 3:0:1:2 sdi 8:128 [active][ghost]
>          \_ 3:0:3:2 sdo 8:224 [active][ghost]
>
>
> dh_state values.
>
>         #cat /sys/block/sdf/device/dh_state
>         rdac
>
>         #cat /sys/block/sdl/device/dh_state
>         rdac
>
>         #cat /sys/block/sdi/device/dh_state
>         rdac
>
>         # cat /sys/block/sdo/device/dh_state
>         rdac
>
>  Please let me know what you think.
>  Note (Kernel version 2.6.27-rc7 and multipath-tools v0.4.8)
>
> Thanks
> Babu Moger
>
> -----Original Message-----
> From: Moger, Babu
> Sent: Friday, October 03, 2008 10:35 PM
> To: sekharan@us.ibm.com
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: RE: [dm-devel] failover does not work with rdac device handler
>
> Thanks Chandra for the response..
>
> That is the good news. So, we do not have to change the kernel  code everytime we want to support the new product.
>
> I will do the steps you mentioned first thing on Monday and reply your back.
>
> Thanks
> Babu Moger
>
> ________________________________________
> From: Chandra Seetharaman [sekharan@us.ibm.com]
> Sent: Friday, October 03, 2008 7:50 PM
> To: Moger, Babu
> Cc: device-mapper development; linux-scsi@vger.kernel.org
> Subject: Re: [dm-devel] failover does not work with rdac device handler
>
> It is supposed to work. I will look into it and respond.
>
> Can you try one thing
>
> for all your devices, do
>         echo rdac > /sys/block/sd?/dh_state
>
> (first you can read the files and see if they all show rdac or not)
>
> and then do your fail over. Let me know what happens.
>
> chandra
> On Fri, 2008-10-03 at 18:12 -0600, Moger, Babu wrote:
> > Hi,
> >
> > I have quick question.
> >
> > I am trying to configure device mapper for my FC array (on my Linux box). I have passed all the arguments from /etc/multipath.conf(like vendor, model etc ..).
> >
> > Discovery seems fine. But, failover is not working.  I am seeing the following error.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > Is it required to add vendor id and model id anywhere else to support this device?
> >
> > Thanks
> > Babu Moger
> >
> >
> > -----Original Message-----
> > From: dm-devel-bounces@redhat.com [mailto:dm-devel-bounces@redhat.com] On Behalf Of Moger, Babu
> > Sent: Wednesday, October 01, 2008 4:48 PM
> > To: device-mapper development; linux-scsi@vger.kernel.org
> > Subject: [dm-devel] failover does not work with rdac device handler
> >
> > Hi All,
> >
> > I am new to the device mapper and just thought I will set-up device mapper on our arrays. I am using the latest kernel 2.6.27.rc6 and got the latest multipath-tools from upstream.
> >
> > I updated /etc/multipath.conf with all the information (vendor, product, hardware_handler, prio etc  ...). Note that my device is not part of rdac_dev_list.
> >
> > Discovery worked fine. "multipath -ll" displays all the information. I can see that it is picking up rdac device handler.
> >
> > But I found that for some reason failover/failback is not working.  I am seeing following these messages.
> >
> > "device-mapper: multipath: Cannot failover device because scsi_dh_rdac was not loaded."
> >
> > I can see that scsi_dh_rdac driver is already loaded.
> >
> > I did some more investigation. Found that failover works fine if I add vendor, model in rdac_dev_list (in file scsi_dh_rdac.c).
> >
> > My question - is it required to add all the supported vendor and model names in rdac_dev_list?
> >
> > Appreciate your responses.
> >
> > Thanks
> > Babu Moger
> >
> >
> >
> >
> >
> >
> >
> > --
> > dm-devel mailing list
> > dm-devel@redhat.com
> > https://www.redhat.com/mailman/listinfo/dm-devel
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html


--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

RE: [dm-devel] failover does not work with rdac device handler

[Chandra Seetharaman]

Hi,

Can you try the attached patch.

chandra
PS: I see a problem (not related to this patch) that the dh_state file
is not getting recreated, still working on it.

-------------

Keep a reference count of attaches, so that same number of detaches are allowed.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
---
Index: linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
===================================================================
--- linux-2.6.27-rc8-git5.orig/drivers/scsi/device_handler/scsi_dh.c
+++ linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
@@ -153,12 +153,26 @@ static int scsi_dh_handler_attach(struct
 	if (sdev->scsi_dh_data) {
 		if (sdev->scsi_dh_data->scsi_dh != scsi_dh)
 			err = -EBUSY;
-	} else if (scsi_dh->attach)
+		else
+			kref_get(&sdev->scsi_dh_data.kref);
+	} else if (scsi_dh->attach) {
 		err = scsi_dh->attach(sdev);
+		if (!err)
+			kref_init(&sdev->scsi_dh_data.kref);
+	}
 
 	return err;
 }
 
+static void scsi_dh_release(struct *kref kref)
+{
+	struct scsi_dh_data *scsi_dh_data;
+	scsi_dh_data = container_of(kref, struct scsi_dh_data, kref);
+
+	if (scsi_dh_data->scsi_dh && scsi_dh_data->scsi_dh->detach)
+		scsi_dh_data->scsi_dh->detach(sdev);
+}
+
 /*
  * scsi_dh_handler_detach - Detach a device handler from a device
  * @sdev - SCSI device the device handler should be detached from
@@ -176,11 +190,7 @@ static void scsi_dh_handler_detach(struc
 	if (scsi_dh && scsi_dh != sdev->scsi_dh_data->scsi_dh)
 		return;
 
-	if (!scsi_dh)
-		scsi_dh = sdev->scsi_dh_data->scsi_dh;
-
-	if (scsi_dh && scsi_dh->detach)
-		scsi_dh->detach(sdev);
+	kref_put(&sdev->scsi_dh_data.kref, scsi_dh_release);
 }
 
 /*
Index: linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
===================================================================
--- linux-2.6.27-rc8-git5.orig/include/scsi/scsi_device.h
+++ linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
@@ -191,6 +191,7 @@ struct scsi_device_handler {
 
 struct scsi_dh_data {
 	struct scsi_device_handler *scsi_dh;
+	struct kref kref;
 	char buf[0];
 };
 

RE: failover does not work with rdac device handler

[Moger, Babu]

Hi Chandra,

I have both good news and bad news with this patch.

Good news is failover is happening (I am not seeing the earlier message "Cannot failover device because scsi_dh_rdac was not loaded").

Bad news is, I am seeing the below soft-panic.

  Oct  7 12:50:15 localhost kernel: BUG: unable to handle kernel NULL pointer dereference at 0000000000000238
Oct  7 12:50:15 localhost kernel: IP: [<ffffffffa038e283>] rdac_bus_detach+0xd/0x9a [scsi_dh_rdac]
Oct  7 12:50:15 localhost kernel: PGD 0
Oct  7 12:50:15 localhost kernel: Oops: 0000 [1] SMP
Oct  7 12:50:15 localhost kernel: CPU 3
Oct  7 12:50:15 localhost kernel: Modules linked in: ipt_REJECT xt_tcpudp iptable_filter ip_tables x_tables bridge stp autofs4 i2c_dev i2c_core hidp rfcomm l2cap bluetooth sunrpc dm_round_robin scsi_dh_rdac dm_multipath scsi_dh sbs sbshc battery acpi_memhotplug ac ipv6 parport_pc lp parport joydev sg bnx2 ide_cd_mod cdrom button rtc_cmos dcdbas serio_raw rtc_core i5000_edac shpchp edac_core rtc_lib pcspkr dm_snapshot dm_zero dm_mirror dm_log dm_mod lpfc qla2xxx scsi_transport_fc ata_piix libata megaraid_sas sd_mod scsi_mod ext3 jbd uhci_hcd ohci_hcd ehci_hcd [last unloaded: microcode]
Oct  7 12:50:15 localhost kernel: Pid: 683, comm: fc_wq_3 Not tainted 2.6.27-rc7-babu #2
Oct  7 12:50:15 localhost kernel: RIP: 0010:[<ffffffffa038e283>]  [<ffffffffa038e283>] rdac_bus_detach+0xd/0x9a [scsi_dh_rdac]
Oct  7 12:50:15 localhost kernel: RSP: 0018:ffff88007d4b3cd0  EFLAGS: 00010282
Oct  7 12:50:15 localhost kernel: RAX: 0000000000000000 RBX: ffff88007d4b3730 RCX: ffffffff00007530
Oct  7 12:50:15 localhost kernel: RDX: 0000000000000000 RSI: 0000000000000040 RDI: ffff88007d4b3730
Oct  7 12:50:15 localhost kernel: RBP: ffffffffa0376000 R08: ffff88007d54ef50 R09: ffff88007fb79090
Oct  7 12:50:15 localhost kernel: R10: 0000000000000000 R11: ffffffff802fa1f5 R12: 0000000000000002
Oct  7 12:50:15 localhost kernel: R13: ffff88007db97920 R14: 0000000000000002 R15: ffff88007db97920
Oct  7 12:50:15 localhost kernel: FS:  0000000000000000(0000) GS:ffff88007f005640(0000) knlGS:0000000000000000
Oct  7 12:50:15 localhost kernel: CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
Oct  7 12:50:15 localhost kernel: CR2: 0000000000000238 CR3: 0000000000201000 CR4: 00000000000006e0
Oct  7 12:50:15 localhost kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Oct  7 12:50:15 localhost kernel: DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Oct  7 12:50:15 localhost kernel: Process fc_wq_3 (pid: 683, threadinfo ffff88007d4b2000, task ffff88007f1f88d0)
Oct  7 12:50:15 localhost kernel: Stack:  ffff88007db97920 ffff88007b0e1c08 ffffffffa0376000 ffffffffa0376029
Oct  7 12:50:15 localhost kernel:  ffff88007d54ef50 ffff88007c45f4b0 ffff88007b0e1c00 ffffffff8031e383
Oct  7 12:50:15 localhost kernel:  ffff88007db97800 ffff88007db97800 ffff88007db97920 ffffffffa0376443
Oct  7 12:50:15 localhost kernel: Call Trace:
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0376000>] ? scsi_dh_release+0x0/0x2e [scsi_dh]
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0376029>] ? scsi_dh_release+0x29/0x2e [scsi_dh]
Oct  7 12:50:15 localhost kernel:  [<ffffffff8031e383>] ? kref_put+0x41/0x4c
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0376443>] ? scsi_dh_notifier+0x73/0x7a [scsi_dh]
Oct  7 12:50:15 localhost kernel:  [<ffffffff80483885>] ? notifier_call_chain+0x29/0x4c
Oct  7 12:50:15 localhost kernel:  [<ffffffff80249a4c>] ? __blocking_notifier_call_chain+0x48/0x62
Oct  7 12:50:15 localhost kernel:  [<ffffffff80396099>] ? device_del+0x150/0x178
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0074f2e>] ? __scsi_remove_device+0x3a/0x7a [scsi_mod]
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0074f8f>] ? scsi_remove_device+0x21/0x2e [scsi_mod]
Oct  7 12:50:15 localhost kernel:  [<ffffffffa0075021>] ? __scsi_remove_target+0x85/0xc3 [scsi_mod]
Oct  7 12:50:15 localhost kernel:  [<ffffffffa00750a4>] ? __remove_child+0x0/0x1a [scsi_mod]
Oct  7 12:50:16 localhost kernel:  [<ffffffffa00750ba>] ? __remove_child+0x16/0x1a [scsi_mod]
Oct  7 12:50:16 localhost kernel:  [<ffffffff80395955>] ? device_for_each_child+0x22/0x4d
Oct  7 12:50:16 localhost kernel:  [<ffffffffa0075099>] ? scsi_remove_target+0x3a/0x45 [scsi_mod]
Oct  7 12:50:16 localhost kernel:  [<ffffffffa00fd7b9>] ? fc_starget_delete+0x0/0x64 [scsi_transport_fc]
Oct  7 12:50:16 localhost kernel:  [<ffffffff802439e4>] ? run_workqueue+0x7a/0x102
Oct  7 12:50:16 localhost kernel:  [<ffffffff802442dc>] ? worker_thread+0xd5/0xe0
Oct  7 12:50:16 localhost kernel:  [<ffffffff80246a8e>] ? autoremove_wake_function+0x0/0x2e
Oct  7 12:50:16 localhost kernel:  [<ffffffff80244207>] ? worker_thread+0x0/0xe0
Oct  7 12:50:16 localhost multipathd: mpath2: load table [0 20971520 multipath 0 1 rdac 2 1 round-robin 0 2 1 8:80 300 8:176 300 round-robin 0 1 1 8:224 100]
Oct  7 12:50:16 localhost kernel:  [<ffffffff80246960>] ? kthread+0x47/0x73
Oct  7 12:50:16 localhost kernel:  [<ffffffff80230fc3>] ? schedule_tail+0x28/0x60
Oct  7 12:50:16 localhost kernel:  [<ffffffff8020cd49>] ? child_rip+0xa/0x11
Oct  7 12:50:16 localhost kernel:  [<ffffffff80246919>] ? kthread+0x0/0x73
Oct  7 12:50:16 localhost kernel:  [<ffffffff8020cd3f>] ? child_rip+0x0/0x11
Oct  7 12:50:16 localhost kernel:
Oct  7 12:50:16 localhost kernel:
Oct  7 12:50:16 localhost kernel: Code: a0 31 c0 e8 8e 80 ea df c6 43 0d 02 eb 08 a8 01 74 04 c6 43 0d 01 5b 5d 44 89 e0 41 5c c3 55 53 48 89 fb 48 83 ec 08 48 8b 47 08 <48> 8b b8 38 02 00 00 e8 e0 30 0f e0 48 89 c6 48 8b 43 08 48 8b
Oct  7 12:50:16 localhost kernel: RIP  [<ffffffffa038e283>] rdac_bus_detach+0xd/0x9a [scsi_dh_rdac]
Oct  7 12:50:16 localhost kernel:  RSP <ffff88007d4b3cd0>
Oct  7 12:50:16 localhost kernel: CR2: 0000000000000238
Oct  7 12:50:16 localhost kernel: ---[ end trace 9be4f9e6c2a759cf ]---

-------------------------------------------------------------------------

Your patch did not compile straight away (on scsi_dh.c). I had to do some changes. I did few changes with the knowledge I had. Correct me if there is anything wrong.
Here is the patch.


--- scsi_dh.c   2008-10-07 10:25:40.000000000 -0500
+++ linux-2.6.27-rc7-babu/drivers/scsi/device_handler/scsi_dh.c 2008-10-07 12:13:19.000000000 -0500
@@ -148,17 +148,35 @@
 static int scsi_dh_handler_attach(struct scsi_device *sdev,
                                  struct scsi_device_handler *scsi_dh)
 {
+
        int err = 0;

-       if (sdev->scsi_dh_data) {
+       if (sdev->scsi_dh_data){
                if (sdev->scsi_dh_data->scsi_dh != scsi_dh)
                        err = -EBUSY;
-       } else if (scsi_dh->attach)
-               err = scsi_dh->attach(sdev);
+               else
+                       kref_get (&sdev->scsi_dh_data->kref);
+       } else if (scsi_dh->attach){
+               err = scsi_dh->attach (sdev);
+               if (!err)
+                       kref_init (&sdev->scsi_dh_data->kref);
+       }

        return err;
 }

+static void scsi_dh_release(struct kref *kref)
+{
+       struct scsi_dh_data *scsi_dh_data;
+       struct scsi_device *sdev;
+       scsi_dh_data = container_of(kref, struct scsi_dh_data, kref);
+       sdev = container_of(&scsi_dh_data, struct scsi_device, scsi_dh_data);
+
+       if (scsi_dh_data->scsi_dh && scsi_dh_data->scsi_dh->detach)
+               scsi_dh_data->scsi_dh->detach(sdev);
+}
+
+
 /*
  * scsi_dh_handler_detach - Detach a device handler from a device
  * @sdev - SCSI device the device handler should be detached from
@@ -175,12 +193,7 @@

        if (scsi_dh && scsi_dh != sdev->scsi_dh_data->scsi_dh)
                return;
-
-       if (!scsi_dh)
-               scsi_dh = sdev->scsi_dh_data->scsi_dh;
-
-       if (scsi_dh && scsi_dh->detach)
-               scsi_dh->detach(sdev);
+       kref_put(&sdev->scsi_dh_data->kref, scsi_dh_release);
 }

 /*



PS:  Yes. You are right. With linux-2.6.27-rc8 sources, I am not seeing the dh_state in sysfs filesystem. That is the reason I reverted back to linux-2.6.27-rc7.

Thanks
Babu Moger

-----Original Message-----
From: Chandra Seetharaman [mailto:sekharan@us.ibm.com]
Sent: Monday, October 06, 2008 9:05 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler

Hi,

Can you try the attached patch.

chandra
PS: I see a problem (not related to this patch) that the dh_state file
is not getting recreated, still working on it.

-------------

Keep a reference count of attaches, so that same number of detaches are allowed.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
---
Index: linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
===================================================================
--- linux-2.6.27-rc8-git5.orig/drivers/scsi/device_handler/scsi_dh.c
+++ linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
@@ -153,12 +153,26 @@ static int scsi_dh_handler_attach(struct
        if (sdev->scsi_dh_data) {
                if (sdev->scsi_dh_data->scsi_dh != scsi_dh)
                        err = -EBUSY;
-       } else if (scsi_dh->attach)
+               else
+                       kref_get(&sdev->scsi_dh_data.kref);
+       } else if (scsi_dh->attach) {
                err = scsi_dh->attach(sdev);
+               if (!err)
+                       kref_init(&sdev->scsi_dh_data.kref);
+       }

        return err;
 }

+static void scsi_dh_release(struct *kref kref)
+{
+       struct scsi_dh_data *scsi_dh_data;
+       scsi_dh_data = container_of(kref, struct scsi_dh_data, kref);
+
+       if (scsi_dh_data->scsi_dh && scsi_dh_data->scsi_dh->detach)
+               scsi_dh_data->scsi_dh->detach(sdev);
+}
+
 /*
  * scsi_dh_handler_detach - Detach a device handler from a device
  * @sdev - SCSI device the device handler should be detached from
@@ -176,11 +190,7 @@ static void scsi_dh_handler_detach(struc
        if (scsi_dh && scsi_dh != sdev->scsi_dh_data->scsi_dh)
                return;

-       if (!scsi_dh)
-               scsi_dh = sdev->scsi_dh_data->scsi_dh;
-
-       if (scsi_dh && scsi_dh->detach)
-               scsi_dh->detach(sdev);
+       kref_put(&sdev->scsi_dh_data.kref, scsi_dh_release);
 }

 /*
Index: linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
===================================================================
--- linux-2.6.27-rc8-git5.orig/include/scsi/scsi_device.h
+++ linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
@@ -191,6 +191,7 @@ struct scsi_device_handler {

 struct scsi_dh_data {
        struct scsi_device_handler *scsi_dh;
+       struct kref kref;
        char buf[0];
 };

RE: [dm-devel] failover does not work with rdac device handler

[Chandra Seetharaman]

Sorry, I sent an imcomplete patch. Here is the correct one.

BTW, The panic you saw is caused by the line (one of the lines you
added):

dev = container_of(&scsi_dh_data, struct scsi_device, scsi_dh_data);

in scsi_dh_release(). We cannot use pointer in a structure to get the 
parent structure.

chandra

Keep a reference count of attaches, so that same number of detaches are allowed.
-----------

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
---
Index: linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
===================================================================
--- linux-2.6.27-rc8-git5.orig/drivers/scsi/device_handler/scsi_dh.c
+++ linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
@@ -153,12 +153,28 @@ static int scsi_dh_handler_attach(struct
 	if (sdev->scsi_dh_data) {
 		if (sdev->scsi_dh_data->scsi_dh != scsi_dh)
 			err = -EBUSY;
-	} else if (scsi_dh->attach)
+		else
+			kref_get(&sdev->scsi_dh_data->kref);
+	} else if (scsi_dh->attach) {
 		err = scsi_dh->attach(sdev);
+		if (!err) {
+			kref_init(&sdev->scsi_dh_data->kref);
+			sdev->scsi_dh_data->sdev = sdev;
+		}
+	}
 
 	return err;
 }
 
+static void scsi_dh_release(struct kref *kref)
+{
+	struct scsi_dh_data *scsi_dh_data;
+	scsi_dh_data = container_of(kref, struct scsi_dh_data, kref);
+
+	if (scsi_dh_data->scsi_dh && scsi_dh_data->scsi_dh->detach)
+		scsi_dh_data->scsi_dh->detach(scsi_dh_data->sdev);
+}
+
 /*
  * scsi_dh_handler_detach - Detach a device handler from a device
  * @sdev - SCSI device the device handler should be detached from
@@ -176,11 +192,7 @@ static void scsi_dh_handler_detach(struc
 	if (scsi_dh && scsi_dh != sdev->scsi_dh_data->scsi_dh)
 		return;
 
-	if (!scsi_dh)
-		scsi_dh = sdev->scsi_dh_data->scsi_dh;
-
-	if (scsi_dh && scsi_dh->detach)
-		scsi_dh->detach(sdev);
+	kref_put(&sdev->scsi_dh_data->kref, scsi_dh_release);
 }
 
 /*
Index: linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
===================================================================
--- linux-2.6.27-rc8-git5.orig/include/scsi/scsi_device.h
+++ linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
@@ -6,6 +6,7 @@
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
 #include <linux/blkdev.h>
+#include <linux/kref.h>
 #include <scsi/scsi.h>
 #include <asm/atomic.h>
 
@@ -191,6 +192,8 @@ struct scsi_device_handler {
 
 struct scsi_dh_data {
 	struct scsi_device_handler *scsi_dh;
+	struct kref kref;
+	struct scsi_device *sdev; /* back reference */
 	char buf[0];
 };
 

RE: failover does not work with rdac device handler

[Moger, Babu]

Hi Chandra,

Yes.  It works fine now.  I tested failover/failback few times and it works as expected. Thanks you very much for helping me to resolve this problem. Please let us know when this change is available in upstream kernel.

Thanks
Babu Moger

-----Original Message-----
From: Chandra Seetharaman [mailto:sekharan@us.ibm.com]
Sent: Tuesday, October 07, 2008 2:19 PM
To: Moger, Babu
Cc: device-mapper development; linux-scsi@vger.kernel.org
Subject: RE: [dm-devel] failover does not work with rdac device handler


Sorry, I sent an imcomplete patch. Here is the correct one.

BTW, The panic you saw is caused by the line (one of the lines you
added):

dev = container_of(&scsi_dh_data, struct scsi_device, scsi_dh_data);

in scsi_dh_release(). We cannot use pointer in a structure to get the
parent structure.

chandra

Keep a reference count of attaches, so that same number of detaches are allowed.
-----------

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
---
Index: linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
===================================================================
--- linux-2.6.27-rc8-git5.orig/drivers/scsi/device_handler/scsi_dh.c
+++ linux-2.6.27-rc8-git5/drivers/scsi/device_handler/scsi_dh.c
@@ -153,12 +153,28 @@ static int scsi_dh_handler_attach(struct
        if (sdev->scsi_dh_data) {
                if (sdev->scsi_dh_data->scsi_dh != scsi_dh)
                        err = -EBUSY;
-       } else if (scsi_dh->attach)
+               else
+                       kref_get(&sdev->scsi_dh_data->kref);
+       } else if (scsi_dh->attach) {
                err = scsi_dh->attach(sdev);
+               if (!err) {
+                       kref_init(&sdev->scsi_dh_data->kref);
+                       sdev->scsi_dh_data->sdev = sdev;
+               }
+       }

        return err;
 }

+static void scsi_dh_release(struct kref *kref)
+{
+       struct scsi_dh_data *scsi_dh_data;
+       scsi_dh_data = container_of(kref, struct scsi_dh_data, kref);
+
+       if (scsi_dh_data->scsi_dh && scsi_dh_data->scsi_dh->detach)
+               scsi_dh_data->scsi_dh->detach(scsi_dh_data->sdev);
+}
+
 /*
  * scsi_dh_handler_detach - Detach a device handler from a device
  * @sdev - SCSI device the device handler should be detached from
@@ -176,11 +192,7 @@ static void scsi_dh_handler_detach(struc
        if (scsi_dh && scsi_dh != sdev->scsi_dh_data->scsi_dh)
                return;

-       if (!scsi_dh)
-               scsi_dh = sdev->scsi_dh_data->scsi_dh;
-
-       if (scsi_dh && scsi_dh->detach)
-               scsi_dh->detach(sdev);
+       kref_put(&sdev->scsi_dh_data->kref, scsi_dh_release);
 }

 /*
Index: linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
===================================================================
--- linux-2.6.27-rc8-git5.orig/include/scsi/scsi_device.h
+++ linux-2.6.27-rc8-git5/include/scsi/scsi_device.h
@@ -6,6 +6,7 @@
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
 #include <linux/blkdev.h>
+#include <linux/kref.h>
 #include <scsi/scsi.h>
 #include <asm/atomic.h>

@@ -191,6 +192,8 @@ struct scsi_device_handler {

 struct scsi_dh_data {
        struct scsi_device_handler *scsi_dh;
+       struct kref kref;
+       struct scsi_device *sdev; /* back reference */
        char buf[0];
 };