phoenix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sergey Soldatov <sergeysolda...@gmail.com>
Subject Re: drop phoenix index cause high cpu
Date Thu, 23 Feb 2017 01:12:06 GMT
You said that data was still writing. Does that mean that the IndexTool was
running when you dropped index? When you create index with ASYNC keyword no
data is written to the index table until you run the IndexTool. Dropping
index that was not loaded by IndexTool should not cause any work on the
region servers.

Thanks,
Sergey

On Mon, Feb 20, 2017 at 6:59 PM, 金砖 <jinzhuan@wacai.com> wrote:

> Hi,
>
> I have a big table with 10Million rows and 60fields.  This table already
> created 8 indexes,  and yesterday I created another index.
>
> *create index table_another_index on table(another_field) async.*
>
> right after this index was created I realized is not needed. So I droppped
> it.
>
> *drop index table_another_index on table; *
> data was stilling writing in the mean time.
>
> *Here comes the problem:*
> load of every regionserver node rised up from 1~2 to 50.  cpu user time up
> to 80-90. this phenomenon continued 5minutes.
>
> *Question:*
> * What happened after async index was dropped ?*
> * I am running phoenix-4.8.0-1.0-hbase with hbase 1.1.3.*
>
>
> *PS:*
>     some exception loged in regionserver log:
> * ERROR [B.defaultRpcServer.handler=253,queue=19,port=16020]
> parallel.BaseTaskRunner: Found a failed task because:
> org.apache.phoenix.hbase.index.exception.SingleIndexWriteFailureException:
> Failed 1 action: Table ’TABLE_ANOTHER_INDEX' was not found, got:
> TABLE_INDEX.: 1 time,*
> *java.util.concurrent.ExecutionException:
> org.apache.phoenix.hbase.index.exception.SingleIndexWriteFailureException:
> Failed 1 action: Table ’TABLE_ANOTHER_INDEX' was not found, got:
> TABLE_INDEX.: 1 time*
>
>
>
> some exception loged in client:
> *o.a.p.execute.MutationState - Swallowing exception and retrying after
> clearing meta cache on connection. java.sql.SQLException: ERROR 2008
> (INT10): Unable to find cached index metadata.  ERROR 2008 (INT10): ERROR
> 2008 (INT10): Unable to find cached index metadata.
>  key=-615404062862804744
> region=TABLE,xxx,1476675558366.cf334587bcdc120cf27e00d5d44ca0a6. Index
> update failed*
>
>
>

Mime
View raw message