On 01/05/2015 10:04 AM, David Fetter wrote:
> On Mon, Jan 05, 2015 at 09:23:47AM -0500, Steve Singer wrote:
>> On 12/30/2014 07:50 PM, David Fetter wrote:
>>> On Thu, Dec 18, 2014 at 03:40:47PM -0500, Steve Singer wrote:
>>>> On 12/18/2014 12:42 PM, Scott Marlowe wrote:
>>>>
>>>> I have added a patch for this against
>>>>
>>>> http://bugs.slony.info/bugzilla/show_bug.cgi?id=356
>>>
>>> ETA on 2.2.4?  This seems like a pretty serious data loss bug and a
>>> pretty small patch to fix it.
>>>
>>
>> I'd like to see us aim for a 2.2 release maybe next week that includes fixes
>> for bug 356,354  and also removes the 'unsupported warning' for 9.4. Bugs
>> 338 and 345 are already committed on the 2.2 branch and would be included as
>> well.
>
> Thanks for the information.
>
>> A few weeks ago I said I wanted to do a release close to the end of
>> the year but that didn't happen.  I make no promises with the above,
>> other people might have other ideas.
>
> What other stakeholders are involved, and do you have any ideas about
> how they might weigh in?

I still haven't gotten a reply to my review request from Jan or Chris on 
those patches.  Also a decision on what todo about bug 350 (see below)

Other than that it involves me actually having the time to package up a 
candidate tar and run tests against various PG versions (mostly 
automated) and to manually test that I can still build against windows 
(not automated).

Any new problems that come up during the above could delay things.



>
>> If the patch (suitable for 2.2) for bug 350 is ready in the next few
>> days I'd say we should include it.  Otherwise maybe we want to
>> clarify the upgrade documentation to better describe the existing
>> behaviour.
>
> 350 is well above my experience level to fix, but I'd be happy to
> document the issue if that's what's needed.

I thought Rose had sent a patch in for this that needed to be tested, 
but I don't see the patch attached to the bug.

The issue here is that the instructions for upgrading to 2.2 say change 
cleanup interval to a few seconds  (
http://www.slony.info/documentation/2.2.2/slonyupgrade.html#AEN2512 )
This doesn't actually work as advertised.

If we aren't going to make cleanup_interval do what something useful 
then we need to provide upgrade instructions work as described. Manually 
forcing a log switch by calling the stored function might be one option.




>
> Cheers,
> David.
>