Brad Hall wrote:
> On Thu, Sep 25, 2008 at 01:20:26PM -0700, Dan Price wrote:
>> On Wed 24 Sep 2008 at 05:00PM, Brad Hall wrote:
>>> On Wed, Sep 24, 2008 at 03:27:49PM -0700, Danek Duvall wrote:
>>>> On Tue, Sep 23, 2008 at 03:36:51PM -0700, Brad Hall wrote:
>>>>
>>>>> http://cr.opensolaris.org/~bhall/bug-388-2/
>> I'm sorry to be late on comments about this RFE.  As I was falling
>> asleep last night it occurred to me that we should probably take
>> a ZFS snapshot (via the BE stuff) if we decide that any fixing is
>> needed.  This could be a boon if the fix goes wrong somehow, or if
>> the user wants to i.e. go back to the pre-fixed state.
> 
> That's a good point -- I'll add support for that.
>  
>> The other thing that occurred to me was that fix might not work
>> properly for certain cases-- such as the libc case, where we mount
>> the processor specific libc over top of the real libc... there must
>> be code in 'verify' that copes with this somehow, but I'm not clear
>> where that code actually lives.
> 
> Not currently sure what to do in that case; but yes, we should handle it.
> 
> Thanks,
> Brad
> _______________________________________________
> pkg-discuss mailing list
> [email protected]
> http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

OK, new webrev posted:
http://cr.opensolaris.org/~bhall/bug-388-4/

This one contains the following changes:
- If it is going to repair anything it takes a zfs snapshot
- Moved to using imageplan rather than calling the actions install 
method directly (per Bart's suggestion); this allows for the fixing of 
hardlinks correctly using the code he already had in there for updates.
- Removed the action::repair functions since the actions sent back by 
verify are added to the update list for the packageplans now, which 
calls install anyways.

Thanks,
Brad
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to