| From: | Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> |
|---|---|
| To: | Stephen Frost <sfrost(at)snowman(dot)net> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: MultiXactId error after upgrade to 9.3.4 |
| Date: | 2014-03-31 13:57:29 |
| Message-ID: | 20140331135729.GW9567@eldon.alvh.no-ip.org |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Stephen Frost wrote:
> Further review leads me to notice that both HEAP_XMAX_IS_MULTI and
> HEAP_XMAX_INVALID are set:
>
> t_infomask | 6528
>
> 6528 decimal -> 0x1980
>
> 0001 1001 1000 0000
>
> Which gives us:
>
> 0000 0000 1000 0000 - HEAP_XMAX_LOCK_ONLY
> 0000 0001 0000 0000 - HEAP_XMIN_COMMITTED
> 0000 1000 0000 0000 - HEAP_XMAX_INVALID
> 0001 0000 0000 0000 - HEAP_XMAX_IS_MULTI
>
> Which shows that both HEAP_XMAX_INVALID and HEAP_XMAX_IS_MULTI are set.
> Of some interest is that HEAP_XMAX_LOCK_ONLY is also set..
This combination seems reasonable. This tuple had two FOR SHARE
lockers, so it was marked HEAP_XMAX_SHARED_LOCK|HEAP_XMAX_IS_MULTI
(0x1080). Then those lockers finished, and somebody else checked the
tuple with a tqual.c routine (say HeapTupleSatisfiesUpdate), which saw
the lockers were gone and marked it as HEAP_XMAX_INVALID (0x800),
without removing the Xmax value and without removing the other bits.
This is all per spec, so we must cope.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Alvaro Herrera | 2014-03-31 14:02:45 | Re: MultiXactId error after upgrade to 9.3.4 |
| Previous Message | Thom Brown | 2014-03-31 13:43:32 | Re: B-Tree support function number 3 (strxfrm() optimization) |