When a new collection is created, Avalon creates a sub-directory with the name of that collection (substituting underscores for any blanks) within the Avalon dropbox directory. A Batch Ingest is initiated by uploading an Ingest Package to the collection sub-directory in the dropbox. To connect to the Avalon dropbox, see Uploading Content to an Avalon Dropbox.
The Manifest File is a spreadsheet (xls, xlsx, csv, or ods) containing the metadata for the items to be created, as well as the file paths to the content files that make up each item; download manifest_example.xlsx for a blank manifest template. The Manifest File lists many metadata fields, but only a few of these fields are required: "Title," "Date Issued," and "File." However, "Bibliographic ID" may be used in place of "Title" and "Date Issues." For a description of the other fields, as well instructions for adding structure or captions, see Batch Ingest Package Format.
Avalon will check for new spreadsheets once every minute. Upon detection, the system will open the spreadsheet and attempt to validate it for correct information. Spreadsheets may fail validation for the following reasons:
If the user account exists in the Avalon system, but the spreadsheet is invalid for some reason, the user will be emailed a failure message with explaining why the spreadsheet was not accepted; multiple reasons may be listed.
Example error message:
User USER_KEY does not have permission to add items to collection: COLLECTION_NAME
If the user account does not exist in the Avalon system (i.e. username was left blank, entered incorrectly, or no email address is associated with the account), the failure message will be sent to the notification email address for Avalon. The default notification email address is "email@example.com," and can be located in
email: comments: 'firstname.lastname@example.org' notifications: 'email@example.com' support: 'firstname.lastname@example.org'
If your instance of Avalon is running on AWS, default emails will need to be verified with AWS first (by clicking a verify link in a message sent to the email address).
If the spreadsheet passes validation, the user will receive an email stating that their spreadsheet has been accepted:
Your metadata package was validated and no errors were found. Your batch is now queued. Manifest file: example_manifest_file.xlsx
This message does not mean that the ingest was successfully completed, only that the sheet was accepted and the rows have been queued for ingest.
Every fifteen minutes, a task will run to check if a Batch Ingest has finished (all rows have been run and were either ingested or failed due to an error). Upon detecting a finished Batch Ingest with any errors, the user will receive email with a report containing a replay name in the format universally-unique-id_original_filename.
Example replay name:
Example batch ingest report, with reported errors:
Any reported errors will always begin at Row 3 (Row 1 contains username information, Row 2 contains headings). Use the row number to find the object causing an error, and use the error description to correct the object information. Refer to Batch Ingest Package Format for the required syntax for each field.
Every twenty-four hours, a task will check for Batch Ingest processes with no activity since the last check. Activity includes changes to the batch registry (username, collection, etc.) and changes to individual entries (entries succeeding or failing). For processes that have been inactive for 4 days or longer, an alert email will be sent to the notification email address for Avalon for all batches with no activity. This alert is informational only; it does not imply that anything is wrong.
By using the replay name contained in the report email, Avalon will rerun a corrected spreadsheet and ingest any previously failed items. Replaying will also update any ingested and unpublished media objects that have received changes. Please note that published media objects may not be corrected using a replayed Manifest File, so this method will always fail if Avalon is auto-publishing media objects.
To replay a Manifest File:
After Avalon detects the uploaded Manifest File, it will queue the rows for ingest as it normally does. When it reaches a changed row, it will take the following actions:
If necessary, replayed Manifest Files can be stacked, although it isn't advisable to do so. After a Manifest File has been corrected and re-uploaded, a user may make another correction and re-upload immediately again. Doing so will cause a changed media object to be processed twice (once for each replayed spreadsheet).
The data for Batch Ingest is stored within an SQL database. Manifest Files are registered in the BatchRegistries table, and individual rows are stored in the BatchEntries table, linked by a key. Admins can view these tables to understand how the spreadsheets are ingested and how information is stored. See the
db/schema.rb file for more information on these tables.
The different tasks that schedule Batch Ingest can be found in the
config/schedule.rb file, and their times can be altered as desired:
avalon:batch:ingest - checks for new spreadsheets
avalon:batch:ingest_status_check- checks for completed batches, sends report emails
avalon:batch:ingest_stalled_check- checks for stalled batches (no activity in 4 days), sends report emails
After a new spreadsheet is detected and validated, an IngestBatchEntryJob is queued as
:ingest (for those who want to throttle jobs in the queue). If the job queue is lost, every row in BatchEntries with a
Queued will need to be re-added to the job queue. The job is documented and defined in