နားလည်ထားရမယ့် အချက်
Incident response တွင် change freeze၊ user impact၊ timeline နဲ့ hypotheses ကိုခွဲထားရပါတယ်။ Redis `INFO` sections၊ latency tools၊ slow log၊ client connections၊ key samples၊ persistence/replication state ကို low-impact အစဉ်ဖြင့်စစ်ရပါတယ်။ Mitigation နဲ့ root-cause fix ကိုခွဲပြီး data-loss window ကိုမဖုံးကွယ်ရပါ။
လက်တွေ့ scenario နဲ့ ချိတ်ကြည့်မယ်
Scenario: memory 95%၊ evictions 50k/min၊ p99 800ms၊ AOF rewrite၊ replica lag 20s၊ leader restart နောက် sessions ပျောက်သည်။ First 15 minutes actions၊ commands၊ stop conditions၊ traffic shedding၊ failover/restore decision နဲ့ stakeholder update ရေးမယ်။
အတူတူ စမ်းရေးကြည့်မယ်
INFO memory
INFO stats
INFO persistence
INFO replication
SLOWLOG GET 20
LATENCY LATEST
CLIENT LIST
MEMORY STATSLow-impact evidence bundle နဲ့ prioritized incident decision tree ရမည်။၅ မိနစ် စမ်းကြည့်
Incident timeline၊ five hypotheses၊ confirm/refute signal၊ immediate fix နဲ့ prevention actions ရေးပါ။
သတိလေးတစ်ချက်
Pressure အချိန် production ပေါ် `KEYS *`၊ full dump သို့ untested failover မလုပ်ပါနှင့်။ Blast radius တိုးနိုင်ပါတယ်။
Redis — Latency Monitoring — Redis