You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@hbase.apache.org by john smith <js...@gmail.com> on 2010/01/28 07:32:20 UTC

Table size from API

Hi guys,

Is there a way to get the rowcount of the table from java API .. I looked at
HTable and HTableDesc... but I couldn't find it .. It is similar to "count
<tablename>" frm Hbase shell.

Thanks
J-S

Re: Table size from API

Posted by Andrew Purtell <ap...@apache.org>.

Actual row count is not stored anywhere. It must be discovered by a table scan, or
a distributed count (mapreduce job, usually).

   - Andy



----- Original Message ----
> From: john smith <js...@gmail.com>
> To: hbase-user@hadoop.apache.org
> Sent: Thu, January 28, 2010 2:44:16 PM
> Subject: Re: Table size from API
> 
> I think META table stores some info about tables .. is it not the case?
> 
> On Thu, Jan 28, 2010 at 12:10 PM, Dan Washusen wrote:
> 
> > I think the shell just performs a scan and iterates over all the rows,
> > counting as it goes.  You could do the same with the client API...
> >
> > 2010/1/28 john smith 
> >
> > > Hi guys,
> > >
> > > Is there a way to get the rowcount of the table from java API .. I looked
> > > at
> > > HTable and HTableDesc... but I couldn't find it .. It is similar to
> > "count
> > > " frm Hbase shell.
> > >
> > > Thanks
> > > J-S
> > >
> >

Re: Table size from API

Posted by john smith <js...@gmail.com>.

I think META table stores some info about tables .. is it not the case?

On Thu, Jan 28, 2010 at 12:10 PM, Dan Washusen <da...@reactive.org> wrote:

> I think the shell just performs a scan and iterates over all the rows,
> counting as it goes.  You could do the same with the client API...
>
> 2010/1/28 john smith <js...@gmail.com>
>
> > Hi guys,
> >
> > Is there a way to get the rowcount of the table from java API .. I looked
> > at
> > HTable and HTableDesc... but I couldn't find it .. It is similar to
> "count
> > <tablename>" frm Hbase shell.
> >
> > Thanks
> > J-S
> >
>

Re: Table size from API

Posted by Dan Washusen <da...@reactive.org>.

I think the shell just performs a scan and iterates over all the rows,
counting as it goes.  You could do the same with the client API...

2010/1/28 john smith <js...@gmail.com>

> Hi guys,
>
> Is there a way to get the rowcount of the table from java API .. I looked
> at
> HTable and HTableDesc... but I couldn't find it .. It is similar to "count
> <tablename>" frm Hbase shell.
>
> Thanks
> J-S
>

Re: Table size from API

Posted by Harry Chen <ha...@hchen1.com>.

The only way I know how to do row count is to use Scan. If you want additional optimization, try FirstKeyOnlyFilter (see its API JavaDoc).

Here is an example:

 final Scan sizeScan = new Scan().setFilter(new FirstKeyOnlyFilter());
 final ResultScanner scanner = htable.getScanner(sizeScan);
 try {
    int count = 0;
    for (Iterator<?> it = scanner.iterator(); it.hasNext();) {
       it.next();
       count++;
    }
    return count;
 } finally {
    scanner.close();
 }

- HC

On Jan 27, 2010, at 10:32 PM, john smith wrote:

> Hi guys,
> 
> Is there a way to get the rowcount of the table from java API .. I looked at
> HTable and HTableDesc... but I couldn't find it .. It is similar to "count
> <tablename>" frm Hbase shell.
> 
> Thanks
> J-S