Merlin Moncure <mmoncure(at)gmail(dot)com> writes:
> On Fri, Mar 23, 2012 at 11:35 AM, Gerhard Wiesinger <lists(at)wiesinger(dot)com> wrote:
>> So I think XID overflow should be planned for one of the next PostgreSQL
>> releases.
> two mitigating factors:
> 1. read only transactions do not increment xid counter
Yes. Ask your admin what his throughput of *data modifying*
transactions is. I'll bet that number is a few orders of magnitude
smaller again.
> 2. xid wraparound counter is per table.
That unfortunately isn't so, the XID counter is global to an installation.
regards, tom lane