Possible problem with getsplits command in accumulo shell [SEC=UNOFFICIAL]

["McClure, Bruce MR 2" <br...@defence.gov.au>]

UNOFFICIAL
Hi,

Thanks for the quick and insightful response to my previous query.

I have experienced an issue using the getsplits command in the accumulo shell.  It occurs when the tableId of the table happens to be a single letter ("w" for example ) and there are other tables in the system that start with that letter ("w1", "w3" for example).  The result is the output of getsplits is displaying splits for other tables.  This has produced outputs with multiple default tablets for example and has been demonstrated in multiple installations.

Has this issue been addressed since version 1.9.2?

Looking at the current getsplits code, it seems to be scanning a range where the start and end of the range are: <tableId>, <tableId>+'<'.  For our example above this amounts to:
accumulo shell -u <user>
table <my table>
scan -b w -e w<

And this can be shown to pick up tablets from tables w1, w3 etc.

So it doesn't look like it has been fixed, unless the fix is to prevent single letter tableIds in some version > 1.9.2

Is there any corresponding bug in the system for this?  (I could not find one).

Thanks again,

Bruce.

Re: Possible problem with getsplits command in accumulo shell [SEC=UNOFFICIAL]

[Christopher <ct...@apache.org>]

That does seem like a bug. We probably should be scanning between:
<tableId> + ';' and <tableId> + '<' instead of scanning without that
semicolon. Would you be willing to create a new issue or pull request
at https://github.com/apache/accumulo ?

On Wed, Nov 10, 2021 at 10:32 PM McClure, Bruce MR 2
<br...@defence.gov.au> wrote:
>
> UNOFFICIAL
>
> Hi,
>
>
>
> Thanks for the quick and insightful response to my previous query.
>
>
>
> I have experienced an issue using the getsplits command in the accumulo shell.  It occurs when the tableId of the table happens to be a single letter (“w” for example ) and there are other tables in the system that start with that letter (“w1”, “w3” for example).  The result is the output of getsplits is displaying splits for other tables.  This has produced outputs with multiple default tablets for example and has been demonstrated in multiple installations.
>
>
>
> Has this issue been addressed since version 1.9.2?
>
>
>
> Looking at the current getsplits code, it seems to be scanning a range where the start and end of the range are: <tableId>, <tableId>+’<’.  For our example above this amounts to:
>
> accumulo shell –u <user>
>
> table <my table>
>
> scan –b w –e w<
>
>
>
> And this can be shown to pick up tablets from tables w1, w3 etc.
>
>
>
> So it doesn’t look like it has been fixed, unless the fix is to prevent single letter tableIds in some version > 1.9.2
>
>
>
> Is there any corresponding bug in the system for this?  (I could not find one).
>
>
>
> Thanks again,
>
>
>
> Bruce.
>
>