[CentOS] 4.5 upgrades on production servers?

[Steve Thompson]

On Wed, 23 May 2007, Matt Hyclak wrote:

> On Wed, May 23, 2007 at 04:54:12PM -0400, Steve Thompson enlightened us:
>>> I upgraded to 4.5 without any problems, but I did have to account for the
>>> fact that the new kernel can potentially reorder NICs on certain HP and
>>> Dell servers, so all was well after I switched the network cable...
>>> See:  http://kbase.redhat.com/faq/FAQ_46_10460.shtm
>>
>> I wonder whose brilliant idea it was to introduce such a major change in a
>> _patch release_?
>
> This has already been hashed out on the nahant-list. Someone from Redhat
> gave an explanation. I will leave it up to you to decided if the explanation
> is satisfactory to you or not.

Heh. I wasn't subscribed to nahant-list, so I went over there and took a 
look. I am in agreement with those that thought that it was a stupid idea. 
A very stupid idea. An extremely dumb stupid idea. The 4.4 -> 4.5 upgrade 
broke my PE2900 systems, and I _do_ use HWADDR as well as bonding. The 
bonding broke in a very spectacular manner. If that change was necessary, 
it should have been introduced in RHEL5, and not in a patch release. Of 
course, it has happened before, when the DLM locking protocols changed 
between RHEL4 U2 and U3 (was it?), leaving non-functional clusters after 
an up2date.

Steve