Re: BUG #16760: Standby database missed records for at least 1 table

On Tue, 8 Dec 2020 at 21:03, Andriy Bartash <abartash(at)xmatters(dot)com> wrote:

> Hi Euler
>
> I inspected pages where I expected to have my data on broken standby.
>
> What was done:
>
> 1. Identified the page where was my data (on primary cluster)
>
> select ctid,audit_ev_id from audit_evs_all where audit_ev_id between
> 221535154 and 221535162;
>
> and it returned
>
> ctid,audit_ev_id
>
> "(1979,3)",221535154
>
> "(1980,13)",221535155
>
> "(1980,14)",221535156
>
> "(1980,28)",221535157
>
> "(1980,29)",221535158
>
> "(1982,6)",221535159
>
> "(1982,7)",221535160
>
> "(1982,32)",221535161
>
> "(1982,33)",221535162
>
> Found that page # 1982 had 4 records on primary (ev_id 221535159,
> 221535160, 221535161 and 221535162)
>
> If you already have the broken secondary, share the following query
results from both servers (requires extension pageinspect):

SELECT lp, lp_off, lp_flags, lp_len, t_xmin, t_xmax, t_field3, t_ctid,
to_hex(t_infomask) AS infomask, to_hex(t_infomask2) AS infomask2, t_hoff,
t_bits FROM heap_page_items(get_raw_page('audit_evs_all', 1982)) WHERE
t_ctid IN('1982,6)', '(1982,7)', '(1982,32)');

--
Euler Taveira http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services