Redshiftのクラスタを再構築せざるを得なかった話 ref: http://qiita.com/momotasasaki/items/6a5e23b1ccdae79c2ead

file0.txt

Dear Customer,

Our systems have identified a potential hardware issue that may affect your 
Amazon Redshift cluster {your_cluster_name} the {your_region_name} AWS region.

As a precaution, we recommend that you restore a new cluster from a snapshot 
within the next 2-3 days and delete your existing cluster.

This will automatically provision you on healthy hardware and help you 
avoid an unplanned outage. Please confirm receipt of this message.

We apologize for the inconvenience, please let us know if you have any questions or concerns.

Regards,
Redshift Service Team

file1.txt

潜在的なハードウェアの問題を特定したので2~3日以内に新しいクラスタをスナップショットから復元して既存のクラスタを削除してね

file3.txt

# CNAMEの確認
$ dig redshift.production

# td-agent.confの変更
$ vim /etc/td-agent/td-agent.conf

file4.txt

# td-agent再起動
$ sudo service td-agent restart

# ログの確認
$ sudo tail -f /var/log/td-agent/td-agent.log

file5.txt

# 新クラスタの対象のテーブルにcount数があることを確認
production=# select count(*) from table_A;
count
-------
    289(適当です)
(1 row)

file6.txt

UNLOAD ('select * from table_A')
TO 's3://{作ったバケット}/table_A/'
CREDENTIALS 'aws_access_key_id=hogehoge;aws_secret_access_key=hugahuga';

file7.txt

COPY action_logFROM 's3://{作ったバケット}/table_A/'
CREDENTIALS 'aws_access_key_id=hgoehgoe;aws_secret_access_key=hugahuga';

truncate.sql

truncate table_A; 

-- 確認用(0だったらok)
select count(*) from table_A;