Connection charset mojibake corruption

On-call incidentMedium30 min

Your task

Support reports user names and addresses with accented and non-Latin characters now show as garbled mojibake (e.g. 'José' to 'José', '北京' to question marks) for records created in the last 3 hours. Older records are fine. Dashboards: error rate normal, no failed inserts. Context: a service was redeployed 3 hours ago after a connection-library upgrade; the MySQL columns are utf8mb4, but the new driver's default connection charset is latin1 (the old driver defaulted to utf8mb4). How do you triage, stop further corruption, and recover the already-corrupted rows?

It covers

  • Encoding
  • Mojibake
  • Charset
  • Mysql