[Unionfs] corruption of rpm database with unionfs (and ldconfig)

[Urs Beyerle]

Erez Zadok wrote:
> In message <47FF0FE1.3010505 at psi.ch>, Urs Beyerle writes:
>> Hi,
>>
>> I still see the same problem (or a similar one) on Scientific Linux 4 
>> (RHEL Kernel 2.6.9-67.0.7.EL) with latest unionfs-2.3.2_for_2.6.9-final. 
>> Did the fix also went to unionfs-2.3.2_for_2.6.9-final?
>
> Hi Urs,
>
> Done.
>
> Please check unionfs-2.3.3, just released.  I was able to back-port the
> fixes from the later kernels (2.6.23+) to kernels 2.6.9 through 2.6.22.  So
> now even those older-ish kernels have moved to using ->fault instead of
> address_space_ops.
>
> I've tested this as thoroughly as I could, but I'd really appreciate some
> more testing on your end (and anyone else).  Feel free to open bugzilla
> reports with any problems you discover (reports of "it's working better now"
> are accepted too :-)


Hi Erez,

I have done some quick tests with unionfs-2.3.3_for_2.6.9-final and 
indeed no rpm database corruption so far!

However, firefox still doesn't start up with unionfs-2.3.3 (it's working 
with unionfs-1.1.5). But this problem is most probably unrelated to the 
"rpm database corruption" problem. I have to take a closer look before I 
can open a bugzilla report.

Conclusion: "it's working better now" and the "rpm database corruption" 
problem seems to be solved.


Thanks,

	Urs