# select+set_keepalive組合操作引起的數據讀寫錯誤 在高并發編程中，我們必須要使用連接池技術，通過減少建連、拆連次數來提高通訊速度。 錯誤示例代碼： ~~~ local redis = require "resty.redis" local red = redis:new() red:set_timeout(1000) -- 1 sec -- or connect to a unix domain socket file listened -- by a redis server: -- local ok, err = red:connect("unix:/path/to/redis.sock") local ok, err = red:connect("127.0.0.1", 6379) if not ok then ngx.say("failed to connect: ", err) return end ok, err = red:select(1) if not ok then ngx.say("failed to select db: ", err) return end ngx.say("select result: ", ok) ok, err = red:set("dog", "an animal") if not ok then ngx.say("failed to set dog: ", err) return end ngx.say("set result: ", ok) -- put it into the connection pool of size 100, -- with 10 seconds max idle time local ok, err = red:set_keepalive(10000, 100) if not ok then ngx.say("failed to set keepalive: ", err) return end ~~~ 如果單獨執行這個用例，沒有任何問題，用例是成功的。但是這段“沒問題”的代碼，卻導致了詭異的現象。 我們的大部分redis請求的代碼應該是類似這樣的： ~~~ local redis = require "resty.redis" local red = redis:new() red:set_timeout(1000) -- 1 sec -- or connect to a unix domain socket file listened -- by a redis server: -- local ok, err = red:connect("unix:/path/to/redis.sock") local ok, err = red:connect("127.0.0.1", 6379) if not ok then ngx.say("failed to connect: ", err) return end ok, err = red:set("cat", "an animal too") if not ok then ngx.say("failed to set cat: ", err) return end ngx.say("set result: ", ok) -- put it into the connection pool of size 100, -- with 10 seconds max idle time local ok, err = red:set_keepalive(10000, 100) if not ok then ngx.say("failed to set keepalive: ", err) return end ~~~ 這時候第二個示例代碼在生產運行中，會出現cat偶會被寫入到數據庫1上，且幾率大約1%左右。出錯的原因在于錯誤示例代碼使用了select(1)操作，并且使用了長連接，那么她就會潛伏在連接池中。當下一個請求剛好從連接池中把他選出來，又沒有重置select(0)操作，那么后面所有的數據操作就都會默認觸發在數據庫1上了。怎么解決，不用我說了吧？