# taskrunner crontab

# Put new 'pending' statuses to LogETL for each feed (for both PHP & Talend feeds)
*/10 * * * * . $HOME/.bash_profile; /usr/bin/php /mnt/data/php-etl/scripts/RawFileIngestion/schedule_job.php >> /mnt/data/php-etl/logs/raw_file_ingestion_schedule_job.log 2>&1

# Extract statuses from LogETL and put them to SQS for raw file ingestion for PHP feeds
*/15 * * * * . $HOME/.bash_profile; /usr/bin/php /mnt/data/php-etl/scripts/RawFileIngestion/producer.php >> /mnt/data/php-etl/logs/raw_file_ingestion_producer.log 2>&1

# Obtain messages from SQS and perform steps of raw file ingestion to staging_raw_<feed> table
*/2 * * * * . $HOME/.bash_profile; /usr/bin/php /mnt/data/php-etl/scripts/RawFileIngestion/consumer.php >> /mnt/data/php-etl/logs/raw_file_ingestion_consumer.log 2>&1

# Transform and ingest Facebook data from ftp://ftp.theorchard.com/Facebook to fact_social table on both 04 and 07 redshift cluster
* 1 * * *     . $HOME/.bash_profile; /usr/bin/php /mnt/data/php-etl/scripts/FactSocial/import.php >> /mnt/data/php-etl/logs/fact_social_import.log 2>&1

# Run PHP DimRefresh ETLs (https://github.com/theorchard/php-etl/tree/master/src/Orchard/DimRefresh)
0 19 * * *    . $HOME/.bash_profile; /mnt/data/script/export_dimension_data.sh >> /mnt/data/php-etl/logs/export_dimension.log 2>&1
30 19 * * *    . $HOME/.bash_profile; /mnt/data/script/transform_dimension_data.sh >> /mnt/data/php-etl/logs/transform_dimension.log 2>&1
0 0 * * *     . $HOME/.bash_profile; /mnt/data/script/sync_dimension_tables.sh >> /mnt/data/ETL/sync_dimension_tables.log 2>&1

# Reset status to pending for SQS internal error
*/5 * * * * . $HOME/.bash_profile; $HOME/.pyenv/shims/python /mnt/data/python/reset_failed_status.py >> /mnt/data/python/reset.log 2>&1