Joerg Schilling wrote:
> Margot Miller <margot.miller at sun.com> wrote:
>
>
>> All,
>>
>> Let's try to get the focus back to the open issues.
>>
>> It looks like the outstanding issues are now:
>>
>> - How does Solaris auditable policy interact with star/rmt?
>>
>
> I cannot answer that.
>
If this rmt is to replace the default rmt, and also introduces new
security features, then it is mandatory for rmt to conform to the Sun
audit policy. I do not believe this to be waivable.
>
>> - What happens if the system crashes during an incremental
>> star? Are these files automatically cleaned up on reboot?
>>
>
> ufsrestore exists since 27 years. I am not aware that somebody
> asked for a similar feature for the case that the kernel crashes
> during a desaster recovery with ufsrestore.
>
> If there was a need for this kind of support, I am sure Sun did add it
> before, but there is no such support. Why should star get different
> treatement?
>
I am not sure you've answered the question here. Mostly I'm concerned
about the possibility of leaving persistent "turds" in the filesystem,
where they may be harder to find.
I'd be a lot happier if these incremental files were located in /var/run
or some other similar place. (/var/tmp if persistence across a reboot
is required.) I understand that there may be problems in dealing with
filename collisions in such cases. Hopefully a solution to such would
be possible.
-- Garrett
