High error rate on nights storage


Night storage are currently failing, the issue is identified and a fix is in progress

Timeline

  • The table managing our sleepscore overflowed an int32 signed integer foreign key column.
  • An ALTER was started but due to the size of the table we changed strategy and added a new column as BIGINT.
  • The Index on this table was then added.
  • Service was restored once an “ANALYZE” was performed on the table.
  • Backfill process was then started to fill the old values of the new column.

Remediation Actions

  • Monitoring of near depleted int32 columns already exists on primary keys, but monitoring needs to be added on foreign keys.

Incident has been resolved

A fix has been implemented, we're now monitoring