adsj@novozymes.com (Adam =?utf-8?Q?Sj=C3=B8gren?=) writes:
> Here's a statement which currently gives an unexpected chunk error:
> efamroot@kat efam=# SELECT * FROM efam.sendreference WHERE id = '189909908';
> ERROR: unexpected chunk number 0 (expected 1) for toast value 1698936148 in pg_toast_10919630
OK ...
> And when I run the suggested query, I get:
> efamroot@kat efam=# select chunk_id, chunk_seq, ctid, xmin, xmax, length(chunk_data) from
pg_toast.pg_toast_10919630where chunk_id = 1698936148 order by 1,2;
> chunk_id | chunk_seq | ctid | xmin | xmax | length
> ------------+-----------+--------------+------------+------+--------
> 1698936148 | 0 | (52888694,2) | 1511390221 | 0 | 1996
> 1698936148 | 1 | (52888694,4) | 1511390221 | 0 | 1148
> (2 rows)
Hmph. So if you EXPLAIN that query, does it show it's doing it as an
indexscan? I'd expect so, but it's always good to make sure.
Assuming it does say that, then the other test I had in mind would
involve "set enable_indexscan = 0", then repeat the EXPLAIN to make
sure that you now get a seqscan plan (you might need to turn off
enable_bitmapscan too), then do the query again and see whether the
results are the same.
regards, tom lane