နားလည်ထားရမယ့် အချက်
Redis Sentinel က non-clustered Redis deployment တွင် monitoring၊ failure detection၊ leader discovery နဲ့ automated failover ပေးပါတယ်။ Sentinels များ consensus ဖြင့် leader down ဆုံးဖြတ်ကြပြီး replica တစ်ခု promote လုပ်ပါတယ်။ Redis data nodes နှစ်ခုသာမဟုတ်ဘဲ independent failure domains တွင် Sentinel processes အနည်းဆုံးသုံးခုသုံးလေ့ရှိပါတယ်။
လက်တွေ့ scenario နဲ့ ချိတ်ကြည့်မယ်
Leader + two replicas + three Sentinels lab ဆောက်ပြီး client ကို static leader hostname မဟုတ်ဘဲ service name ဖြင့် Sentinel discovery သုံးစေမယ်။ Leader kill test တွင် detection time၊ promotion၊ client reconnect၊ write error window နဲ့ stale old-leader fencing ကိုတိုင်းမယ်။
အတူတူ စမ်းရေးကြည့်မယ်
SENTINEL MASTERS
SENTINEL MASTER tutorial-primary
SENTINEL REPLICAS tutorial-primary
SENTINEL SENTINELS tutorial-primary
SENTINEL GET-MASTER-ADDR-BY-NAME tutorial-primary
SENTINEL CKQUORUM tutorial-primaryMonitored leader၊ replicas၊ peer Sentinels နဲ့ quorum health ရမည်။၅ မိနစ် စမ်းကြည့်
AZ တစ်ခုဆုံးသွားသည့် scenario အတွက် 3/5 Sentinel placement နဲ့ quorum တွက်ပါ။
သတိလေးတစ်ချက်
Sentinels အားလုံး host/AZ တစ်ခုတည်းထားလျှင် shared failure မှာ quorum အကျိုးမရှိပါ။
Redis — Sentinel — Redis