[Bucardo-general] test suite broken on git HEAD?

[Paul Theodoropoulos]

On 9/26/13 1:37 AM, Antoine Pitrou wrote:
> Le Wed, 25 Sep 2013 10:58:16 -0700,
> Paul Theodoropoulos <paul at anastrophe.com> a
> écrit :
>
>> On 9/25/13 3:08 AM, Antoine Pitrou wrote:
>>> #          $got->[1] = ARRAY(0x3218468)
>>> #     $expected->[1] = Does not exist
>>>
>>> [lots of similar failures snipped]
>>>
>>> The Bucardo-4.99.7 test suite works fine in the same context.
>>> Should I stick to Bucardo-4.99.7 for evaluation purposes or is HEAD
>>> actually safe?
>>>
>> I experience exactly the same failures when I run the test suite on
>> the latest git revisions. I've never been able to get the tests to
>> run all the way through to the end without error. What are the
>> particulars of your installation? Perhaps we can find a common
>> problem. I'm on Debian Squeeze, 64bit, perl 5.10, postgresql 8.4.
> Ubuntu 13.04, x86-64, Postgres 9.1, Perl 5.14.
>
>
Hmm, not terribly much in common! Odd. I should add however, that after 
several months of testing (largely learning to understand how all these 
things work together, and fixing countless misconfigurations and 
missed-configurations), I've been doing stress testing on our test setup 
the last two days, with excellent results.

local master in florida, remote master in oregon. Performing thousands, 
and tens of thousands of writes/updates, serially and/or simultaneously, 
breaking the network connection between the servers for long intervals 
(an hour, +), short intervals (simulating a flakey WAN connection, up 
down up down etc), with no errors, and bucardo syncing like a champ. So 
even though I get the above errors on make test, they appear to be 
irrelevant.

-- 
Paul Theodoropoulos
www.anastrophe.com