On Tue, Aug 23, 2011 at 8:49 AM, Daniel Shahaf <d.s_at_daniel.shahaf.name> wrote:
> Hyrum K Wright wrote on Tue, Aug 23, 2011 at 07:52:06 -0500:
>> On Tue, Aug 23, 2011 at 7:06 AM, C. Michael Pilato <cmpilato_at_collab.net> wrote:
>> > On 08/23/2011 06:29 AM, Daniel Shahaf wrote:
>> >> - Bert points out that the HTTPv2 code may not interact well with
>> >> clients committing via the svn.eu mirror to an HTTPv2-enabled svn.us.
>> >
>> > Hrm, I can't think of any scenarios in which having a newer master version
>> > than slave version will cause issues. (It's the reverse case that can be
>> > problematic.) But then, I'm not completely awake yet, either.
>>
>> Although I don't want to be playing fast-and-loose with the ASF repo,
>> running the RC on svn.apache.org would give us just the type of
>> testing in these types of scenarios that's difficult to do manually.
>>
>> Assuming there is a decent "what to do if this thing borks the server"
>> plan, I'm +1.
>>
>
> We'll backup the pre-upgrade data and binaries.
>
> If the post-upgrade data gets corrupted, we'll be depending on the
> corruption not propagating to the mirrors on the 1.6 slave. (and we'll
> have to downtime the us mirror whilst the backups are promoted)
>
> If the upgrade compromises the machine... it's going to be fun.
Heh. Glad to see you've thought about it.
I'd also like to take the opprotunity to remind folks that rc1 isn't
official yet (and there's currently some speculation in IRC that it
might never be), so you may want to hold off upgrading until an
official release is made.
-Hyrum
--
uberSVN: Apache Subversion Made Easy
http://www.uberSVN.com/
Received on 2011-08-23 15:53:53 CEST