Causes of Downtime in MySQL
summary
Analysis of several hundred emergency cases finds three main categories of downtime with most faults outside MySQL in the operating environment; the single top storage cause is disks full, followed by SAN failures due to complexity and poor diagnosis. Performance hits stem from bad SQL, indexing and surprisingly many long-running transactions. Replication is fragile and human errors reflect bad tools or processes; data loss rarely comes from InnoDB. Recommendations: change control, selective monitoring, promiscuous graphing, reboot occasionally, and stop using untested scripts.
AI-written summary · about
