ipoid populates a database via the script init-db.js, which reads a data file that typically has tens of millions of lines.
Inserts may be unsuccessful due to the nature of the data for a particular IP address, e.g.:
- Bad data (e.g. actor_data.org has a character limit, but we have seen instances of a long error message where the organization name should be, which is too long to insert - T325633#8866840)
- Legitimate data that doesn't fit our schema (in which case the problem is with our schema, e.g. see: https://gerrit.wikimedia.org/r/c/mediawiki/services/ipoid/+/921500 )
The script currently crashes if it encounters an error, but instead it should catch and log the error, and continue.