Hi!
As it is seen from the code (toasting.c and further) Toast tables are created immediately
when a new relation with the TOASTable column is created. Practically, there could occur
the case when Toast table does not exist and we should of course check for that.
TOAST_TUPLE_THRESHOLD is not only one which decides should be value stored
externally, this is slightly more complex and less obvious logic:
(see heapam.c, heap_prepare_insert())
else if (HeapTupleHasExternal(tup) || tup->t_len > TOAST_TUPLE_THRESHOLD)
as you can see here is another condition - HeapTupleHasExternal, which is set in
heap_fill_tuple and lower in fill_val, where the infomask bit HEAP_HASEXTERNAL is set.
So when I experimented with the TOAST I'd to add a new flag which forced the value to be
TOASTed regardless of its size.
Also, TOAST_TUPLE_THRESHOLD sets overall tuple size over which it would be considered
to be toasted, and has its minimum value that could not be decreased further.
In [1] (the Pluggable TOAST) we suggest making this an ontion for Toaster.
On Thu, 15 Sept 2022 at 04:04, Aleksander Alekseev
<aleksander@timescale.com> wrote:
> 1. Forbid setting toast_tuple_target < TOAST_TUPLE_THRESHOLD
> 2. Consider using something like RelationGetToastTupleTarget(rel,
> TOAST_TUPLE_THRESHOLD) in heapam.c:2250, heapam.c:3625 and
> rewriteheap.c:636 and modify the documentation accordingly.
> 3. Add a separate user-defined table setting toast_tuple_threshold
> similar to toast_tuple_target.
>
> Thoughts?
There was some discussion on this problem in [1].
The problem with #2 is that if you look at
heapam_relation_needs_toast_table(), it only decides if the toast
table should be created based on (tuple_length >
TOAST_TUPLE_THRESHOLD). So if you were to change the logic as you
describe for #2 then there might not be a toast table during an
INSERT/UPDATE.
The only way to fix that would be to ensure that we reconsider if we
should create a toast table or not when someone changes the
toast_tuple_target reloption. That can't be done under
ShareUpdateExclusiveLock, so we'd need to obtain an
AccessExclusiveLock instead when changing the toast_tuple_target
reloption. That might upset some people.
The general direction of [1] was to just increase the minimum setting
to TOAST_TUPLE_THRESHOLD, but there were some concerns about breaking
pg_dump as we'd have to error if someone does ALTER TABLE to set the
toast_tuple_target reloption lower than the newly defined minimum
value.
I don't quite follow you on #3. If there's no toast table we can't toast.
David
[1] https://www.postgresql.org/message-id/20190403063759.GF3298@paquier.xyz
--
Regards,
Nikita Malakhov
Postgres Professional
