Source
ArctosDB/arctos#9057
Summary
The current procedures were developed under the idea that people would clean up after themselves after we'd dealt with the backlog. That's clearly not true, people are not maintaining their data, and the polite/manual operation is tedious and dangerous (lots of places for me to push the wrong button).
- write a procedure (??) to dump and delete from all tables and stream all to a gzip archive
- rewrite the announce code such that it can be pasted directly into an issue with the data archive
- rewrite the notification to warn of possible data loss (although I suspect scripts are safer than manual, given the scope this has become).
Source
ArctosDB/arctos#9057
Summary
The current procedures were developed under the idea that people would clean up after themselves after we'd dealt with the backlog. That's clearly not true, people are not maintaining their data, and the polite/manual operation is tedious and dangerous (lots of places for me to push the wrong button).