Re: Parallel INSERT (INTO ... SELECT ...)

On Wed, Dec 9, 2020 at 10:11 AM Greg Nancarrow <gregn4422(at)gmail(dot)com> wrote:
>
> On Wed, Dec 9, 2020 at 1:35 AM vignesh C <vignesh21(at)gmail(dot)com> wrote:
> >
> > Most of the code present in
> > v9-0001-Enable-parallel-SELECT-for-INSERT-INTO-.-SELECT.patch is
> > applicable for parallel copy patch also. The patch in this thread
> > handles the check for PROPARALLEL_UNSAFE, we could slightly make it
> > generic by handling like the comments below, that way this parallel
> > safety checks can be used based on the value set in
> > max_parallel_hazard_context. There is nothing wrong with the changes,
> > I'm providing these comments so that this patch can be generalized for
> > parallel checks and the same can also be used by parallel copy.
>
> Hi Vignesh,
>
> You are absolutely right in pointing that out, the code was taking
> short-cuts knowing that for Parallel Insert,
> "max_parallel_hazard_context.max_interesting" had been set to
> PROPARALLEL_UNSAFE, which doesn't allow that code to be generically
> re-used by other callers.
>
> I've attached a new set of patches that includes your suggested improvements.
>

/*
+ * PrepareParallelMode
+ *
+ * Prepare for entering parallel mode, based on command-type.
+ */
+void
+PrepareParallelMode(CmdType commandType)
+{
+ Assert(!IsInParallelMode() || force_parallel_mode != FORCE_PARALLEL_OFF);
+
+ if (IsModifySupportedInParallelMode(commandType))
+ {
+ /*
+ * Prepare for entering parallel mode by assigning a
+ * FullTransactionId, to be included in the transaction state that is
+ * serialized in the parallel DSM.
+ */
+ (void) GetCurrentTransactionId();
+ }
+}

Why do we need to serialize the transaction ID for 0001? I mean in
0001 we are just allowing the SELECT to be executed in parallel so why
we would need the transaction Id for that. I agree that we would need
this once we try to perform the Insert also from the worker in the
remaining patches.

--
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com