[rhel7-branch][master] fix error recovery in custom spoke

[David Lehman]

Still awaiting verification from bug reporter, but I was reproducing
it and this fixed it for me.

[PATCH] Update disk refs when recovering from a devicefactory failure. (#1032141)

[David Lehman]

---
 pyanaconda/ui/gui/spokes/custom.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/pyanaconda/ui/gui/spokes/custom.py b/pyanaconda/ui/gui/spokes/custom.py
index 2f2422a..6ca4a20 100644
--- a/pyanaconda/ui/gui/spokes/custom.py
+++ b/pyanaconda/ui/gui/spokes/custom.py
@@ -1448,6 +1448,12 @@ class CustomPartitioningSpoke(NormalSpoke, StorageChecker):
                     if _device is None:
                         # in this case we have removed the old device so we now have
                         # to re-create it
+
+                        # the disks need to be updated since we've replaced all
+                        # of the devices with copies in the devicefactory error
+                        # handler
+                        old_disk_names = [d.name for d in old_disks]
+                        old_disks = [self.__storage.devicetree.getDeviceByName(n) for n in old_disk_names]
                         try:
                             self._replace_device(old_device_type, device.size,
                                                  disks=old_disks,

[PATCH] Update disk refs when recovering from a devicefactory failure. (#1032141)

[Vratislav Podzimek]

On Mon, 2014-02-24 at 15:42 -0600, David Lehman wrote:
> pyanaconda/ui/gui/spokes/custom.py | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/pyanaconda/ui/gui/spokes/custom.py b/pyanaconda/ui/gui/spokes/custom.py
> index 2f2422a..6ca4a20 100644
> --- a/pyanaconda/ui/gui/spokes/custom.py
> +++ b/pyanaconda/ui/gui/spokes/custom.py
> @@ -1448,6 +1448,12 @@ class CustomPartitioningSpoke(NormalSpoke, StorageChecker):
>                      if _device is None:
>                          # in this case we have removed the old device so we now have
>                          # to re-create it
>                    # the disks need to be updated since we've replaced all
>                    # of the devices with copies in the devicefactory error
>                    # handler
>                    old_disk_names = [d.name for d in old_disks]
>                    old_disks = [self.__storage.devicetree.getDeviceByName(n) for n in old_disk_names]
>                    try:
>                        self._replace_device(old_device_type, device.size,
>                                             disks=old_disks,

Looks good to me.

[PATCH] Update disk refs when recovering from a devicefactory failure. (#1032141)

[Anne Mulhern]

----- Original Message -----
> From: "David Lehman" 
> To: [e-mail]
> Sent: Monday, February 24, 2014 4:42:22 PM
> Subject: [PATCH] Update disk refs when recovering from a devicefactory	failure. (#1032141)
>
> pyanaconda/ui/gui/spokes/custom.py | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/pyanaconda/ui/gui/spokes/custom.py
> b/pyanaconda/ui/gui/spokes/custom.py
> index 2f2422a..6ca4a20 100644
> --- a/pyanaconda/ui/gui/spokes/custom.py
> +++ b/pyanaconda/ui/gui/spokes/custom.py
> @@ -1448,6 +1448,12 @@ class CustomPartitioningSpoke(NormalSpoke,
> StorageChecker):
>                      if _device is None:
>                          # in this case we have removed the old device so we
>                          now have
>                          # to re-create it
>                    # the disks need to be updated since we've replaced
>
> all
>                    # of the devices with copies in the devicefactory
>
> error
>                    # handler
>                    old_disk_names = [d.name for d in old_disks]
>                    old_disks =
>
> [self.__storage.devicetree.getDeviceByName(n) for n in old_disk_names]
>                          try:
>                              self._replace_device(old_device_type,
>                              device.size,
>                                                   disks=old_disks,
> --
> 1.8.5.3
>
> anaconda-patches mailing list
> [e-mail]
> https://lists.fedorahosted.org/mailman/listinfo/anaconda-patches

I recently fixed a bug (#981316) where getDeviceByID() was returning None for disks which were no longer there.
Is it guaranteed that getDeviceByName() will always find a disk?

- mulhern

[PATCH] Update disk refs when recovering from a devicefactory failure. (#1032141)

[David Lehman]

On Tue, 2014-02-25 at 07:34 -0500, Anne Mulhern wrote:
> ----- Original Message -----
> > From: "David Lehman" 
> > To: [e-mail]
> > Sent: Monday, February 24, 2014 4:42:22 PM
> > Subject: [PATCH] Update disk refs when recovering from a devicefactory	failure. (#1032141)
> >
> > pyanaconda/ui/gui/spokes/custom.py | 6 ++++++
> >  1 file changed, 6 insertions(+)
> >
> > diff --git a/pyanaconda/ui/gui/spokes/custom.py
> > b/pyanaconda/ui/gui/spokes/custom.py
> > index 2f2422a..6ca4a20 100644
> > --- a/pyanaconda/ui/gui/spokes/custom.py
> > +++ b/pyanaconda/ui/gui/spokes/custom.py
> > @@ -1448,6 +1448,12 @@ class CustomPartitioningSpoke(NormalSpoke,
> > StorageChecker):
> >                      if _device is None:
> >                          # in this case we have removed the old device so we
> >                          now have
> >                          # to re-create it
> >                    # the disks need to be updated since we've replaced
> >
> > all
> >                    # of the devices with copies in the devicefactory
> >
> > error
> >                    # handler
> >                    old_disk_names = [d.name for d in old_disks]
> >                    old_disks =
> >
> > [self.__storage.devicetree.getDeviceByName(n) for n in old_disk_names]
> >                          try:
> >                              self._replace_device(old_device_type,
> >                              device.size,
> >                                                   disks=old_disks,
> > --
> > 1.8.5.3
> >
> > anaconda-patches mailing list
> > [e-mail]
> > https://lists.fedorahosted.org/mailman/listinfo/anaconda-patches
>
> I recently fixed a bug (#981316) where getDeviceByID() was returning None for disks which were no longer there.
> Is it guaranteed that getDeviceByName() will always find a disk?

It should be, yes. There is no way to change the disk set within the
custom spoke, let alone within a call to Blivet.factoryDevice. Non-disk
block devices come and go, but disk set should always remain constant
within any given visit to the custom spoke.

David
> mulhern
>
> anaconda-patches mailing list
> [e-mail]
> https://lists.fedorahosted.org/mailman/listinfo/anaconda-patches