Re: Need Multixact Freezing Docs

On 04/16/2014 01:30 PM, Alvaro Herrera wrote:
> Josh Berkus wrote:
>>
>>> You can see the current multixact value in pg_controldata output. Keep
>>> timestamped values of that somewhere (a table?) so that you can measure
>>> consumption rate. I don't think we provide SQL-level access to those
>>> values.
>>
>> Bleh. Do we provide SQL-level access in 9.4? If not, I think that's a
>> requirement before release.
>
> Yeah, good idea. Want to propose a patch?

Yeah, lemme dig into this. I really think we need it for 9.4, feature
frozen or not.

>>>> Also: how do I check the multixact age of a table? There doesn't seem
>>>> to be any data for this ...
>>>
>>> pg_class.relminmxid is the oldest multixact value that might be present
>>> in a table.
>>
>> On every database I've tested, age(relminmxid) returns int_max. So this
>> is apparently broken.
>
> Hmm, are you sure it's INT_MAX and not 4244967297? Heikki reported
> that: http://www.postgresql.org/message-id/52401AEA.9000608@vmware.com
> The absolute value is not important; I think that's mostly harmless. I
> don't think applying age() to a multixact value is meaningful, though;
> that's only good for Xids.

Yeah, I'm sure:

josh=# select relname, age(relminmxid) from pg_class;
relname | age
-----------------------------------------+------------
pg_statistic | 2147483647
pg_type | 2147483647
random | 2147483647
dblink_pkey_results | 2147483647
pg_toast_17395 | 2147483647

...

So if age() doesn't mean anything, then how are users to know when the
need to freeze?

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com