phoenix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ankit Singhal <ankitsingha...@gmail.com>
Subject Re: Dropping of Index can still leave some non-replayed writes Phoenix-2915
Date Wed, 15 Jun 2016 15:18:30 GMT
Yes, restart your cluster

On Wed, Jun 15, 2016 at 8:17 AM, anupama agarwal <anu1307@gmail.com> wrote:

> I have created async index with same name. But I am still getting the same
> error. Should I restart my cluster for changes to reflect?
> On Jun 15, 2016 8:38 PM, "Ankit Singhal" <ankitsinghal59@gmail.com> wrote:
>
>> Hi Anupama,
>>
>> Option 1:-
>> You can create a ASYNC index so that WAL can be replayed. And once your
>> regions are up , remember to do the flush of data table before dropping the
>> index.
>>
>> Option 2:-
>> Create a table in hbase with the same name as index table name by using
>> hbase shell.
>>
>> Regards,
>> Ankit Singhal
>>
>>
>> On Tue, Jun 14, 2016 at 11:19 PM, anupama agarwal <anu1307@gmail.com>
>> wrote:
>>
>>> Hi All,
>>>
>>> I have hit this error in phoenix, Phoenix-2915. It could be possible ,
>>> that there are some index writes in WAL which are not replayed and the
>>> index is dropped.
>>>
>>> And, now the table is not there, these writes cannot be replayed which
>>> result in data table regions also to not come up. My data region is in
>>> FAILED_TO_OPEN state. I have tried recreating the index, and still region
>>> is not able to come up. I realise that this has been fixed in new version
>>> of phoenix, but I am currently on phoenix 4.6 and Hbase 1.0. Can you please
>>> suggest a solution?
>>>
>>
>>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message