Table scan/count hangs on particular table

[Rural Hunter <ru...@gmail.com>]

Hi,

We have a hbase 0.98.13-hadoop2 cluster with 7 rs. Recently we found the
plain scan/count shell command hangs on one particular table while other
tables are fine. The java scan api also hangs up if I don't specific a
start row key. But I can scan with some starting row keys of the table. The
get command is fine.

I used hdfs fsck and hbase hbck to check the table and hfiles and they are
all fine. I also found I get 0 row with get shell command using the start
row key of some regions shown in web ui, especially on one rs none of the
start row key of the regions returns any result.

What happened with this table?

Re: Table scan/count hangs on particular table

[Ted Yu <yu...@gmail.com>]

You can narrow the scope of search by issuing scan with start row of region
N and stop row of region N+1 (repeat for regions with #rows > 0).
Suppose you find that scanning region R hangs. You can capture stack trace
on the server which hosts R and pastebin it.

Thanks

On Mon, Dec 21, 2015 at 8:47 PM, Rural Hunter <ru...@gmail.com> wrote:

> Hi,
>
> We have a hbase 0.98.13-hadoop2 cluster with 7 rs. Recently we found the
> plain scan/count shell command hangs on one particular table while other
> tables are fine. The java scan api also hangs up if I don't specific a
> start row key. But I can scan with some starting row keys of the table. The
> get command is fine.
>
> I used hdfs fsck and hbase hbck to check the table and hfiles and they are
> all fine. I also found I get 0 row with get shell command using the start
> row key of some regions shown in web ui, especially on one rs none of the
> start row key of the regions returns any result.
>
> What happened with this table?
>