Re: ERROR: found multixact from before relminmxid

On April 9, 2018 7:51:19 PM PDT, Peter Geoghegan <pg(at)bowt(dot)ie> wrote:
>On Mon, Apr 9, 2018 at 6:56 PM, Alexandre Arruda <adaldeia(at)gmail(dot)com>
>wrote:
>> (... and all other indexes returns null too)
>>
>> I tried with bt_index_check too. Same results.
>
>That's interesting, because it tells me that you have a table that
>appears to not be corrupt, despite the CLUSTER error. Also, the error
>itself comes from sanity checking added to MultiXact freezing fairly
>recently, in commit 699bf7d0.
>
>You didn't say anything about regular VACUUM being broken. Do you find
>that it works without any apparent issue?
>
>I have a suspicion that this could be a subtle bug in
>CLUSTER/freezing. The only heap_freeze_tuple() caller is code used by
>CLUSTER, so it's not that hard to imagine a MultiXact freezing bug
>that is peculiar to CLUSTER. Though I haven't thought about it in much
>detail.

I've not followed this thread. Possible it's the overeager check for pg upgraded tuples from before 9.3 that Alvaro fixed recently?

Andres

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.