问题
To BigQuery experts,
I am working on the process which requires us to represent customers shopping history in way where we concatenate all last 12 months of transactions in a single column for Solr faceting using prefixes.
while trying to load this data in BIG Query, we are getting below row limit exceed error. Is there any way to get around this? the actual tuple size is around 64 mb where as the avro limit is 16mb.
[ ~]$ bq load --source_format=AVRO --allow_quoted_newlines --max_bad_records=10 "syw-dw-prod":"MAP_ETL_STG.mde_golden_tbl" "gs://data/final/tbl1/tbl/part-m-00005.avro"
Waiting on bqjob_r7e84784c187b9a6f_0000015ee7349c47_1 ... (5s) Current status: DONE
BigQuery error in load operation: Error processing job 'syw-dw-prod:bqjob_r7e84784c187b9a6f_0000015ee7349c47_1': Avro parsing error in position 893786302. Size of data
block 27406834 is larger than the maximum allowed value 16777216.
回答1:
BigQuery's limit on loaded Avro file's block size is 16MB (https://cloud.google.com/bigquery/quotas#import). Unless each row is actually greater than 16MB, you should be able to split up the rows into more blocks to stay within the 16MB block limit. Using a compression codec may reduce the block size.
来源:https://stackoverflow.com/questions/46611127/bq-load-error-avro-parsing-error-in-position-893786302-size-of-data-block-274