This GUC hawq_rm_stmt_nvseg is used by resource manager. Yi, could you please give some detailed information?
On Fri, Jul 1, 2016 at 12:48 PM, Vineet Goel <[email protected]> wrote: > What about hawq_rm_stmt_nvseg ? > > > http://hdb.docs.pivotal.io/20/reference/guc/parameter_definitions.html#hawq_rm_stmt_nvseg > > -Vineet > > > > On Jun 30, 2016, at 6:43 PM, Hubert Zhang <[email protected]> wrote: > > You can try to the GUC > > enforce_virtual_segment_number. > > It works for queries without UDF. > > On Fri, Jul 1, 2016 at 9:40 AM, yanqing weng <[email protected]> > wrote: > > > Thanks Hubert. > > > > Can it be set only use one virtual segment for the query ? Because this > > can debug query more easily. > > > > > > On Fri, Jul 1, 2016 at 9:35 AM, Hubert Zhang <[email protected]> wrote: > > > >> For hash tables, The number of virtual segments is equal to the bucket > >> number of the table. Bucket number can be specified when creating the > table. > >> > >> On Fri, Jul 1, 2016 at 9:31 AM, Ivan Weng <[email protected]> wrote: > >> > >>> How about hash table? > >>> > >>> On Fri, Jul 1, 2016 at 9:30 AM, Hubert Zhang <[email protected]> > wrote: > >>> > >>>> For random table, the virtual segment number is related to the data > >>> size of > >>>> the table. For large tables, we prefer use more virtual segments to > >>> achieve > >>>> better performance. While for small tables, less virtual segments are > >>>> needed, which could save resources and enable more concurrent queries. > >>> > >>>> > >>>> On Fri, Jul 1, 2016 at 9:22 AM, yanqing weng <[email protected]> > >>>> wrote: > >>>> > >>>>> Hi HAWQ developer, > >>>>> > >>>>> Elastic execution engine is the important feature in HAWQ 2.0. > >>>>> > >>>>> I have a question for this feature in the HAWQ debugging. Is there > >>>> possible > >>>>> to control the virtual segment number when run query ? If so, which > >>> GUC > >>>> or > >>>>> parameter I should set ? Thanks. > >>>>> > >>>> > >>>> > >>>> > >>>> -- > >>>> Thanks > >>>> > >>>> Hubert Zhang > >>>> > >>> > >> > >> > >> > >> -- > >> Thanks > >> > >> Hubert Zhang > >> > > > > > > > -- > Thanks > > Hubert Zhang > > -- Thanks Hubert Zhang
