Thanks for the review Darren.
On 06/02/11 13:28, Darren Kenny wrote:
Hi Matt,
Generally, looking good, but have some comments, mainly nits:
In auto_install.py:
- You could probably re-write the following code:
641 zpools_to_import = None
642 for zpool in [z for z in desired_zpools if not z.is_root]:
643 if zpools_to_import is None:
644 zpools_to_import = zpool.name
645 else:
646 zpools_to_import = zpools_to_import + " " + zpool.name
as simply:
non_root_zpools = [z.name for z in desired_zpool if not z.is_root]
zpools_to_import = " ".join(non_root_zpools)
This should work too.
Done
- The zp_fh.close() shouldn't be required any more due to the with statement.
In svc-system-config;
- I wonder if the name "zpool" might be too generic, and maybe should be
something more specific like "zpool_imports" or "zpool_config"?
I'm open to any suggestions... at the moment I have
property_groupp = "zpool"
property = "import"
These could be renamed to :
property_group = "zpool_config"
property = "import_pools"
- I'm assuming the need for the -f flag on the import is unavoidable due the
fact that it's possible for a data zpool to have swap or dump devices in it,
is that right?
When an install completes, the data pools will have been created with
the mini-root hostid, when you attempt to import them on first boot the
host id is different and a "-f" is required to ensure the import succeeds.
cheers
Matt
Thanks,
Darren.
On 02/06/2011 12:03, Matt Keenan wrote:
Hi,
Can I get a code review for bug :
7048223 Multi-pool installed data pools need to be imported on first
boot
http://monaco.sfbay.sun.com/detail.jsf?cr=7048223
Webrev:
http://cr.opensolaris.org/~mattman/bug-7048223/
Some background :
With the integration of CUD based AI installations, Multi-Pool
support has been added. All pools other than the root pool (data pools)
created during an AI install do not get automatically imported when the
installed client boots. This is because they are not created within the
newly installed BE.
Not importing these pools on first boot could potentially have
catastrophic effects, in scenarios where dump/swap volumes were being
created on a data pool instead of the root pool, or where user account
home directories are being created on a non root pool, These datasets
would not be available on boot if the pools are not imported.
I've spent considerable time testing various scenarios to try and get
this to happen, and initiated a discussion on zfs-discuss to determine
the easiest means of achieving this.
Usage of zpool create/import with -R altroot and -o cachefile options
were mentioned as potential solutions, however these have proved
unsuccessful, due to hostid issues.
The proposed solution extends the System Config SMF used on first
boot, adding support to import data zpools specified via a profile which
is created during install and placed on the new BE.
cheers
Matt
_______________________________________________
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
