Greenplum failed segment的恢复方法--primary与mirror都可修复

　　当在使用greenplum过程中有不当的操作时，可能会出现segment节点宕掉的情况（比如在greenplum运行的过程中停掉其中几台segment节点的服务器），通过下面的方法可以恢复segment。

下面是现场出现的故障情况：

[gpadmin@tj-soc-c04-csfb1 ~]$ gpstate -m
20161010:16:35:54:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-Starting gpstate with args: -m
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-local Greenplum Version: 'postgres (Greenplum Database) 4.3.6.2 build 1'
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-master Greenplum Version: 'PostgreSQL 8.2.15 (Greenplum Database 4.3.6.2 build 1) on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.4.2 compiled on Nov 12 2015 23:50:28'
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-Obtaining Segment details from master...
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:--------------------------------------------------------------
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:--Current GPDB mirror list and status
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:--Type = Group
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:--------------------------------------------------------------
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-   Mirror             Datadir                       Port    Status              Data Status
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb2   /data1/gpdata/mirror/gpseg0   41000   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb2   /data1/gpdata/mirror/gpseg1   41001   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb3   /data1/gpdata/mirror/gpseg2   41000   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb3   /data1/gpdata/mirror/gpseg3   41001   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-   tj-soc-c04-csfb4   /data1/gpdata/mirror/gpseg4   41000   Acting as Primary   Change Tracking
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:-   tj-soc-c04-csfb4   /data1/gpdata/mirror/gpseg5   41001   Acting as Primary   Change Tracking
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb1   /data1/gpdata/mirror/gpseg6   41000   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-tj-soc-c04-csfb1   /data1/gpdata/mirror/gpseg7   41001   Failed                                <<<<<<<<
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[INFO]:--------------------------------------------------------------
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-2 segment(s) configured as mirror(s) are acting as primaries
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-6 segment(s) configured as mirror(s) have failed
20161010:16:35:55:026100 gpstate:tj-soc-c04-csfb1:gpadmin-[WARNING]:-2 mirror segment(s) acting as primaries are in change tracking

可以看到有6个节点Failed，有2个节点的Primary和Mirror交换了。

一、首先需要停掉GP

gpstop -M fast -a    这样会告诉你有几个节点DOWN了

二、启动GP

gpstart    启动数据库会忽略DOWN的节点

三、生成一个恢复配置文件

gprecoverseg -o ./recov    会在当前目录生成一个recov文件，里面包含了要恢复的节点信息

recov文件内容如下：（注意：这个文件不是手动创建的，而是通过gprecoverseg -o ./recov命令生成的

filespaceOrder=
tj-soc-c04-csfb2:41000:/data1/gpdata/mirror/gpseg0
tj-soc-c04-csfb2:41001:/data1/gpdata/mirror/gpseg1
tj-soc-c04-csfb3:41000:/data1/gpdata/mirror/gpseg2
tj-soc-c04-csfb3:41001:/data1/gpdata/mirror/gpseg3
tj-soc-c04-csfb1:41000:/data1/gpdata/mirror/gpseg6
tj-soc-c04-csfb1:41001:/data1/gpdata/mirror/gpseg7

四、使用恢复配置文件恢复节点

$gprecoverseg -i ./recov

恢复过程中可以用gpstate -m 查看恢复状态：Resynchronizing（表示正在恢复中），Synchronized（表示恢复完毕）

五、调整Primary和Mirror

上面的情况中有Primary和Mirror兑换的情况，所以需要把他们换回来，可以用下面的命令

gprecoverseg -r

等待所有的节点都是Synchronized后，segment就恢复好了