Обсуждение: Why does TRUNCATE require a special privilege?
Hi, From the documentation: > TRUNCATE quickly removes all rows from a set of tables. It has the same > effect as an unqualified DELETE on each table, but since it does not actually > scan the tables it is faster. > (...) > You must have the TRUNCATE privilege on a table to truncate it. Granted that TRUNCATE and DELETE are different operations under the hood, but why would the TRUNCATE operation require its own specific privilege rather than say, use the same privilege as the DELETE operation? Any explanations or lore about this would be deeply appreciated. Thank you, Marcelo.
On Fri, Jan 16, 2026 at 10:13 AM Marcelo Fernandes <marcefern7@gmail.com> wrote: > From the documentation: > > TRUNCATE quickly removes all rows from a set of tables. It has the same > > effect as an unqualified DELETE on each table, but since it does not actually > > scan the tables it is faster. > > (...) > > You must have the TRUNCATE privilege on a table to truncate it. > > Granted that TRUNCATE and DELETE are different operations under the hood, but > why would the TRUNCATE operation require its own specific privilege rather than > say, use the same privilege as the DELETE operation? It's kinda obvious, when you read the notes. 1) Not MVCC-safe. 2) Do not fire TRIGGERs, thus breaking data-integrity 3) "Viral" in the presence of FKs, i.e. related tables must also be TRUNCATEd Just these 3 are HUGE departures from a DELETE. --DD
On 1/16/26 02:32, Dominique Devienne wrote: > On Fri, Jan 16, 2026 at 10:13 AM Marcelo Fernandes <marcefern7@gmail.com> wrote: >> From the documentation: >>> TRUNCATE quickly removes all rows from a set of tables. It has the same >>> effect as an unqualified DELETE on each table, but since it does not actually >>> scan the tables it is faster. >>> (...) >>> You must have the TRUNCATE privilege on a table to truncate it. >> >> Granted that TRUNCATE and DELETE are different operations under the hood, but >> why would the TRUNCATE operation require its own specific privilege rather than >> say, use the same privilege as the DELETE operation? > > It's kinda obvious, when you read the notes. > > 1) Not MVCC-safe. > 2) Do not fire TRIGGERs, thus breaking data-integrity It will not fire ON DELETE triggers, it will fire ON TRUNCATE triggers. > 3) "Viral" in the presence of FKs, i.e. related tables must also be TRUNCATEd Only if you add the CASCADE option, or TRUNCATE them in the same command. Otherwise it will fail. > > Just these 3 are HUGE departures from a DELETE. --DD I would add from: https://www.postgresql.org/docs/current/sql-truncate.html "TRUNCATE acquires an ACCESS EXCLUSIVE lock on each table it operates on, which blocks all other concurrent operations on the table. When RESTART IDENTITY is specified, any sequences that are to be restarted are likewise locked exclusively. If concurrent access to a table is required, then the DELETE command should be used instead." and "" When RESTART IDENTITY is specified, the implied ALTER SEQUENCE RESTART operations are also done transactionally; that is, they will be rolled back if the surrounding transaction does not commit. Be aware that if any additional sequence operations are done on the restarted sequences before the transaction rolls back, the effects of these operations on the sequences will be rolled back, but not their effects on currval(); that is, after the transaction currval() will continue to reflect the last sequence value obtained inside the failed transaction, even though the sequence itself may no longer be consistent with that. This is similar to the usual behavior of currval() after a failed transaction." > > -- Adrian Klaver adrian.klaver@aklaver.com
Those operations do different things, sure. But from a roles/privilege framework perspective, why would you want to give certain users the DELETE privilege whereas others you want to give them only the TRUNCATE privilege? Are we saying to a user that "You need a different level of privilege because you are about to cause a MVCC-unsafe operation?". Or is the privilege framework simply ruling "Do different things, have different permissions"? Marcelo. On Sat, Jan 17, 2026 at 5:46 AM Adrian Klaver <adrian.klaver@aklaver.com> wrote: > > On 1/16/26 02:32, Dominique Devienne wrote: > > On Fri, Jan 16, 2026 at 10:13 AM Marcelo Fernandes <marcefern7@gmail.com> wrote: > >> From the documentation: > >>> TRUNCATE quickly removes all rows from a set of tables. It has the same > >>> effect as an unqualified DELETE on each table, but since it does not actually > >>> scan the tables it is faster. > >>> (...) > >>> You must have the TRUNCATE privilege on a table to truncate it. > >> > >> Granted that TRUNCATE and DELETE are different operations under the hood, but > >> why would the TRUNCATE operation require its own specific privilege rather than > >> say, use the same privilege as the DELETE operation? > > > > It's kinda obvious, when you read the notes. > > > > 1) Not MVCC-safe. > > 2) Do not fire TRIGGERs, thus breaking data-integrity > > It will not fire ON DELETE triggers, it will fire ON TRUNCATE triggers. > > > 3) "Viral" in the presence of FKs, i.e. related tables must also be TRUNCATEd > > Only if you add the CASCADE option, or TRUNCATE them in the same > command. Otherwise it will fail. > > > > > Just these 3 are HUGE departures from a DELETE. --DD > > I would add from: > > https://www.postgresql.org/docs/current/sql-truncate.html > > "TRUNCATE acquires an ACCESS EXCLUSIVE lock on each table it operates > on, which blocks all other concurrent operations on the table. When > RESTART IDENTITY is specified, any sequences that are to be restarted > are likewise locked exclusively. If concurrent access to a table is > required, then the DELETE command should be used instead." > > and > > "" > When RESTART IDENTITY is specified, the implied ALTER SEQUENCE RESTART > operations are also done transactionally; that is, they will be rolled > back if the surrounding transaction does not commit. Be aware that if > any additional sequence operations are done on the restarted sequences > before the transaction rolls back, the effects of these operations on > the sequences will be rolled back, but not their effects on currval(); > that is, after the transaction currval() will continue to reflect the > last sequence value obtained inside the failed transaction, even though > the sequence itself may no longer be consistent with that. This is > similar to the usual behavior of currval() after a failed transaction." > > > > > > > > > -- > Adrian Klaver > adrian.klaver@aklaver.com
On 1/16/26 13:15, Marcelo Fernandes wrote:
> Those operations do different things, sure.
>
> But from a roles/privilege framework perspective, why would you want to give
> certain users the DELETE privilege whereas others you want to give them
> only the TRUNCATE privilege?
>
> Are we saying to a user that "You need a different level of privilege because
> you are about to cause a MVCC-unsafe operation?".
The thing I see as difference worthy of separation is:
"CASCADE
Automatically truncate all tables that have foreign-key references
to any of the named tables, or to any tables added to the group due to
CASCADE.
"
Now the same end result can be done with:
delete from some_table:
where the FK's pointing at some_table have ON CASCADE DELETE. The
difference being a DBA has the option of creating the FKs with ON
CASCADE NO ACTION which would throw an error. In other words you can
prevent an unconstrained DELETE on some_table from removing all the
child records. With TRUNCATE ... CASCADE, you cannot, the potential for
harm is greater.
>
> Or is the privilege framework simply ruling "Do different things, have
> different permissions"?
>
> Marcelo.
>
--
Adrian Klaver
adrian.klaver@aklaver.com
Marcelo Fernandes <marcefern7@gmail.com> writes:
> But from a roles/privilege framework perspective, why would you want to give
> certain users the DELETE privilege whereas others you want to give them
> only the TRUNCATE privilege?
> Are we saying to a user that "You need a different level of privilege because
> you are about to cause a MVCC-unsafe operation?".
Personally I think that's a plenty good enough reason ;-).
The very different locking level is another good reason.
TRUNCATE will block all other activity on the table, while
DELETE doesn't.
However, looking at our git history, it appears that TRUNCATE was
originally only permitted to the table owner. The separate permission
bit was added (years later) so that the owner could grant out the
ability to others, without doing anything as non-backwards-compatible
as reinterpreting what operations a DELETE grant allows.
regards, tom lane