Unloading a large dataset from Redshift to S3 fails with an I / O error

What I'm really trying to accomplish is changing the data types for multiple columns. In particular, several columns that are stored as rows that must be dates. I tried to use the CREATE TABLE / SELECT command with similar problems. I / O errors or broken connections.

I realized that it would be more efficient to export and re-import the data than trying to abuse the cluster by reading / writing so much data in a circle more or less.

I tried both Aginy Workbench and SQL Workbench J with large timeouts (10 hours). SQL Workbench J was able to run for 2 hours and 12 minutes today before failing with the same error that I continue to see again and again.

An I / O error occurred while sending to the backend.

This is a fair piece of data ... 2,028,448,405 rows at present (I say "now" because we add about 70 million rows per day). But I would expect Redshift to handle this easily.

UNLOAD ( 'select weekday, day, month, year, guid,

      ...,

      colN
  from actions a
  where a.colN in (\ 'my \', \ 'penguin \', \ 'lolz \') '
)
TO 's3: //penguin-lolz-bucket/all.csv' 
CREDENTIALS 'aws_access_key_id = $ id; aws_secret_access_key = $ key
ALLOWOVERWRITE; </code>

- .

1:

, 3 :

  • TO
  • GZIP PARALLEL

, / .

2: . ( , 1)

PRO TIP. , AWS Redshift ( "", ). SQL Workbench J. 5 SQL Workbench J , . AWS Redshift , , .

+4
1

, , SQL Workbench .

script PHP shell ( pgsql_connect) .

, script . script , , .

: EC2 script EC2.

+1

All Articles