Re: Sequence moves forward when failover is triggerred

Craig Ringer wrote:
> On 07/11/2012 07:23 AM, Andy Chambers wrote:
>>
>> I think I made a poor decision by having our application
>> generate checkbook numbers on demand using sequences.
>
> Sure did. Sequences are exempt from most transactional rules;
> that's why they're fast and lock-free.
>

This may be another case for the "gap-less" sequence (I hate that
term ... please let's call it a "uniformly-increasing sequence").

http://archives.postgresql.org/pgsql-general/2006-08/msg00744.php