Importing large dumps

Importing large dumps

DBNova was built around one question: can it import a production dump without falling over? A 2 GB .sql file with 20 million rows imports in about two minutes with flat worker memory.

#Supported files

.sql, .sql.gz and .zip (containing one .sql file), up to 100 GB by default.

#The import wizard

  1. Choose a file.
  2. Upload. The file is sent in chunks. If the connection drops, the upload resumes where it stopped.
  3. Choose the target database.
  4. Preflight. DBNova streams through the file and rejects truncated dumps, dumps that touch several databases and files that aren't SQL dumps, before anything reaches MySQL.
  5. Import. A queued job pipes the file into the mysql client.
  6. Result. Errors are reported at the line they occur in your file, with the MySQL message.

#Honest progress

The status card shows "Sent to MySQL", "Statements processed", "Elapsed" and "Errors". It reports bytes sent, never a made-up "% committed".

#Cancel and recovery

  • Cancel stops the mysql client cleanly.
  • If the worker dies, the scheduler marks the import as interrupted within about two minutes.
  • Recovery is drilled: a killed worker, a stopped MySQL server, a full disk and a dropped upload connection all end in a clear state.

Warning For dumps you don't fully trust, sign in with an account that only has rights on the target database. The one-database check is a best-effort guard; MySQL grants are the real boundary.

0 views Something unclear? Open an issue on GitHub