I agree with Niall - the changes he highlighted below
look like a workaround, rather than a fix. I'm wondering
if GUI also now needs to do something similar?
As I'm sure you know, selecting a VTOC disk in TC with
use_whole_disk=True doesn't really use the whole disk - it
applies a default layout and sets Disk.whole_disk=False,
while it does something different for GPT.
So, Disk.whole_disk isn't meant to be a reliable indicator
of whether the disk was previously selected in "use whole disk"
mode.
- Dermot
On 04/26/12 04:01, Niall Power wrote:
Hey Drew,
Thanks for fixing these.
I know this has been reviewed and integrated and I would have given it
a +1 too, but I want to point out that I think we have a problem with
target controller if he have to do this:
--- a/usr/src/cmd/text-install/fdisk_partitions.py Tue Apr 24
11:31:32 2012 -0600
+++ b/usr/src/cmd/text-install/fdisk_partitions.py Wed Apr 25
09:16:11 2012 -0600
@@ -263,6 +263,7 @@
LOGGER.debug("Setting whole_disk for %s", self.disk)
self.disk = self.tc.select_disk(self.disk,
use_whole_disk=True)[0]
+ self.disk.whole_disk = True
else:
# it's a partition, set the in_zpool attribute in the
object
# for now. The next screen will fill in needed slices
diff -r 15ba29ecb4ed usr/src/cmd/text-install/gpt_partitions.py
--- a/usr/src/cmd/text-install/gpt_partitions.py Tue Apr 24
11:31:32 2012 -0600
+++ b/usr/src/cmd/text-install/gpt_partitions.py Wed Apr 25
09:16:11 2012 -0600
@@ -178,6 +178,7 @@
LOGGER.debug("Setting whole_disk for %s", self.disk)
self.use_whole_segment = True
self.disk = self.tc.select_disk(self.disk,
use_whole_disk=True)[0]
+ self.disk.whole_disk = True
else:
LOGGER.debug("Setting whole_disk to false")
self.use_whole_segment = False
So we tell target controller to select the disk with
use_whole_disk=True, and then have to follow up afterwards by manually
setting disk.whole_disk to True.
I'd argue this is more of a workaround than anything. Why should I
have to tell the disk it is in whole_disk mode when I have already
asked target controller to set that up with a whole_disk layout for me?
Of course the whole_disk flag is a bit misleading when VTOC and FDISK
are involved vs. GPT. There is higher risk in fixing this in target
controller but that's where the long term fix belongs IMO.
Cheers,
Niall
On 04/26/12 02:08 AM, Drew Fisher wrote:
Good morning!
Could I get a couple of eyes on:
https://cr.opensolaris.org/action/browse/caiman/drewfish/ti_bugs/webrev/
7120827 <http://monaco.us.oracle.com/detail.jsf?cr=7120827> TI should
check whether the current solaris2 partition size larger the
recommended minimum size 7164145
<http://monaco.us.oracle.com/detail.jsf?cr=7164145> text-install
choose "Use the whole disk" can not jump to "Network" screen directly
7120820 <http://monaco.us.oracle.com/detail.jsf?cr=7120820> TI should
report error if 2 solaris2 partitions existing on the system to be
installed
I've tested each of these locally on my test system. Since they're
each checking for "negative" situations, there's no need to do full
installs.
Thanks!
-Drew
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
