On Wed, 23 Dec 2020 at 4:15 AM, Tomas Vondra
wrote:
> On 11/11/20 8:52 PM, Tomas Vondra wrote:
> > Hi,
> >
> > I took a look at this today, doing a bit of stress-testing, and I can
> > get it to crash because of segfaults in pagetable_create (not sure if
> > the issue is there, it might be just a
On 11/11/20 8:52 PM, Tomas Vondra wrote:
> Hi,
>
> I took a look at this today, doing a bit of stress-testing, and I can
> get it to crash because of segfaults in pagetable_create (not sure if
> the issue is there, it might be just a symptom of an issue elsewhere).
>
> Attached is a shell script
Hi,
I took a look at this today, doing a bit of stress-testing, and I can
get it to crash because of segfaults in pagetable_create (not sure if
the issue is there, it might be just a symptom of an issue elsewhere).
Attached is a shell script I use to run the stress test - it's using
'test' databa
> BTREE support it), then we will create a parallel bitmap heap scan
> path on top of the parallel bitmap index scan path. So the idea of
> this patch is that each worker will do the parallel index scan and
> generate their part of the bitmap. And, we will create a barrier so
>
nt time in preparing the bitmap. So the
>
> > idea of this patch is to use the parallel index scan for preparing the
>
> > underlying bitmap in parallel.
>
> >
>
> > Design:
>
> > If underlying index AM supports the parallel path (currently only
>
> >
> BTREE support it), then we will create a parallel bitmap heap scan
> path on top of the parallel bitmap index scan path. So the idea of
> this patch is that each worker will do the parallel index scan and
> generate their part of the bitmap. And, we will create a barrier so
>
e
> > > bitmap index scan path.
>
> Workers Planned: 4
> -> Parallel Bitmap Heap Scan on tenk1
> Recheck Cond: (hundred > 1)
> - -> Bitmap Index Scan on tenk1_hundred
> + -> Parallel B
e
> > > bitmap index scan path.
>
> Workers Planned: 4
> -> Parallel Bitmap Heap Scan on tenk1
> Recheck Cond: (hundred > 1)
> - -> Bitmap Index Scan on tenk1_hundred
> + -> Parallel B
llel Bitmap Heap Scan on tenk1
Recheck Cond: (hundred > 1)
- -> Bitmap Index Scan on tenk1_hundred
+ -> Parallel Bitmap Index Scan on tenk1_hundred
Index Cond: (hundred > 1)
+1,
> BTREE support it), then we will create a parallel bitmap heap scan
> path on top of the parallel bitmap index scan path. So the idea of
> this patch is that each worker will do the parallel index scan and
> generate their part of the bitmap. And, we will create a barrier so
>
patch is to use the parallel index scan for preparing the
underlying bitmap in parallel.
Design:
If underlying index AM supports the parallel path (currently only
BTREE support it), then we will create a parallel bitmap heap scan
path on top of the parallel bitmap index scan path. So the idea of
11 matches
Mail list logo
