Home | About | Sematext search-lucene.com search-hadoop.com
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB
 Search Hadoop and all its subprojects:

Switch to Threaded View
Pig >> mail # user >> What is the best way to do counting in pig?


Copy link to this message
-
Re: What is the best way to do counting in pig?
instead of doing "dump relation," do "explain relation" (then run
identically) and paste the output here. It will show whether the combiner
is being used,

2012/7/3 Ruslan Al-Fakikh <[EMAIL PROTECTED]>

> Hi,
>
> As it was said, COUNT is algebraic and should be fast, because it
> forces combiner. You should make sure that combiner is really used
> here. It can be disabled in some situations. I've encountered such
> situations many times when a job is tooo heavy in case no combiner is
> applied.
>
> Ruslan
>
> On Tue, Jul 3, 2012 at 1:35 AM, Subir S <[EMAIL PROTECTED]> wrote:
> > Right!!
> >
> > Since it is mentioned that job is hanging, wild guess is it must be
> > 'group all'. How can that be confirmed?
> >
> > On 7/3/12, Jonathan Coveney <[EMAIL PROTECTED]> wrote:
> >> group all uses a single reducer, but COUNT is algebraic, and as such,
> will
> >> use combiners, so it is generally quite fast.
> >>
> >> 2012/7/2 Subir S <[EMAIL PROTECTED]>
> >>
> >>> Group all - uses single reducer AFAIU. You can try to count per group
> >>> and sum may be.
> >>>
> >>> You may also try with COUNT_STAR to include NULL fields.
> >>>
> >>> On 7/3/12, Sheng Guo <[EMAIL PROTECTED]> wrote:
> >>> > Hi all,
> >>> >
> >>> > I used to use the following pig script to do the counting of the
> >>> > records.
> >>> >
> >>> > m_skill_group = group m_skills_filter by member_id;
> >>> > grpd = group m_skill_group all;
> >>> > cnt = foreach grpd generate COUNT(m_skill_group);
> >>> >
> >>> > cnt_filter = limit cnt 10;
> >>> > dump cnt_filter;
> >>> >
> >>> >
> >>> > but sometimes, when the records get larger, it takes lots of time and
> >>> hang
> >>> > up, and or die.
> >>> > I thought counting should be simple enough, so what is the best way
> to
> >>> do a
> >>> > counting in pig?
> >>> >
> >>> > Thanks!
> >>> >
> >>> > Sheng
> >>> >
> >>>
> >>
>
>
>
> --
> Best Regards,
> Ruslan Al-Fakikh
>
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB