Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Accumulo, mail # user - Two locations for the same extent [SEC=UNOFFICIAL]


Copy link to this message
-
Re: Two locations for the same extent [SEC=UNOFFICIAL]
Keith Turner 2013-08-15, 05:51
I guess the master can not scan the metadata table because an execption is
thrown when it sees those two entries.  So you will probably need to remove
the entries from the metadata table.
On Thu, Aug 15, 2013 at 1:12 AM, Dickson, Matt MR <
[EMAIL PROTECTED]> wrote:

> **
>
> *UNOFFICIAL*
> We've killed the servers and they are showing up dead in the Accumulo
> tablet server status screen.  Should the METADATA table entries for the
> extent change now that those servers are offline?  I'm still seeing the two
> extents mapping to the ip's of the dead servers.
>
> Interestingly, when we go to the 'Tables' view on the console there are no
> statistics for any of the tables.
>  ------------------------------
> *From:* Keith Turner [mailto:[EMAIL PROTECTED]]
> *Sent:* Thursday, 15 August 2013 14:49
>
> *To:* [EMAIL PROTECTED]
> *Subject:* Re: Two locations for the same extent [SEC=UNOFFICIAL]
>
>  Killing the two tablet servers may clear it up.
>
>
> On Thu, Aug 15, 2013 at 12:39 AM, Dickson, Matt MR <
> [EMAIL PROTECTED]> wrote:
>
>> **
>>
>> *UNOFFICIAL*
>> We are running 1.5.1.  I found the duplicate definition in the !METADATA
>> table, should it be deteled from there?
>>
>>  ------------------------------
>> *From:* Keith Turner [mailto:[EMAIL PROTECTED]]
>> *Sent:* Thursday, 15 August 2013 13:23
>> *To:* [EMAIL PROTECTED]
>> *Subject:* Re: Two locations for the same extent [SEC=UNOFFICIAL]
>>
>>   what version of Accumulo are you running?
>>
>>
>> On Wed, Aug 14, 2013 at 9:07 PM, Dickson, Matt MR <
>> [EMAIL PROTECTED]> wrote:
>>
>>> **
>>>
>>> *UNOFFICIAL*
>>>
>>> Hi,
>>>
>>> We have had several tablet servers die recently while doing splits and
>>> since restarting we are getting a 'found two locations for the same extent
>>> ....' .  I noticed ACCUMULO-1243 addresses the bug that creates this,
>>> however I just wanted to know how I can fix the two locations for the
>>> extent.  Do I delete one of them or merge them?  How do I identify the
>>> extents?
>>>
>>> Thanks in advance,
>>> Matt Dickson
>>>
>>>
>>
>