記一次hadoop大數(shù)據(jù)集群生產(chǎn)事故

陸續(xù)對(duì)原有的hadoop、hbase集群做了擴(kuò)容，增加了幾個(gè)節(jié)點(diǎn)，中間沒(méi)有重啟過(guò)，今天早上發(fā)現(xiàn)一個(gè)hregionserver服務(wù)停止了，就先啟動(dòng)服務(wù)，沒(méi)想到啟動(dòng)之后一直有訪問(wèn)數(shù)據(jù)的出錯(cuò)，嘗試對(duì)整個(gè)hbase集群進(jìn)行重啟出現(xiàn)了下面的錯(cuò)誤：

$ start-hbase.sh

master running as process 112580. Stop it first.

The authenticity of host 'szc-l0104567 (192.168.1.81)' can't be established.

RSA key fingerprint is 76:e5:12:90:de:59:e1:da:02:f3:f1:2a:9a:a6:f8:c4.

Are you sure you want to continue connecting (yes/no)? The authenticity of host 'szc-l0104566 (192.168.1.80)' can't be established.

RSA key fingerprint is cd:d1:ad:98:ca:36:b5:ec:c3:1d:be:b8:8c:ae:bc:80.

Are you sure you want to continue connecting (yes/no)? The authenticity of host 'szc-l0124500 (192.168.1.95)' can't be established.

RSA key fingerprint is ec:3e:83:b0:bf:f0:3b:6d:7e:fa:e8:1d:7e:67:ed:27.

通過(guò)字面意思可以看出是驗(yàn)證出現(xiàn)問(wèn)題，在驗(yàn)證主機(jī)名無(wú)密碼通過(guò)的時(shí)候需要通過(guò)輸入“yes”才能進(jìn)行下去。

我是通過(guò)拷貝authorized_keys文件到對(duì)方節(jié)點(diǎn)的.ssh目錄進(jìn)行無(wú)密碼通過(guò)的，這樣只能通過(guò)針對(duì)IP的無(wú)密碼驗(yàn)證，不需要輸入“yes”，但是在第一次無(wú)密碼驗(yàn)證主機(jī)名的時(shí)候還需要輸入“yes”，所以在添加hregionser節(jié)點(diǎn)的時(shí)候，無(wú)密碼驗(yàn)證就需要手動(dòng)的去做第一次的主機(jī)名無(wú)密碼驗(yàn)證，或者用下面的方式做無(wú)密碼驗(yàn)證。

方式2：

ssh-copy-id -i ~/.ssh/id_rsa.pub username@IP

ssh-copy-id -i ~/.ssh/id_rsa.pub username@HOSTNAME

方式3：

在執(zhí)行完第一步拷貝authorized_keys或者ssh-copy-id -i ~/.ssh/id_rsa.pub username@IP 之后再執(zhí)行下面的步驟

ssh -o stricthostkeychecking=no HOSTNAME