数据库内核月报－ 2016/04 - GPDB · 特性分析 · Segment 修复指南 - 《数据库内核月报》

问题背景
GP的Segment修复
实现原理

问题背景

GPDB是中央控制节点式的架构，在一个 GreenPlum 集群中，有一个 Master 节点和多个 Segment 节点。Master 是中央控制节点，Segment 是数据存放节点。所有的Segment节点平等，均由Master管理。架构如下图：

GreenPlum架构图

当GP Master出现问题的时候，可以通过外部的HA监控模块发现并激活备库，Standby Master 正常后删除原来的 Master 进行重建备库。

而 Segment 的修复与此不同！由上图可知，Segment 也分为主备，称为 Primary 和 Mirror，Mirror 是 Primary 的备。Primary与Mirror之间强同步保证数据一致性和可靠性，其间的监控与切换则由Master的FTS模块负责。当FTS发现Primary宕机、Mirror健康后会激活Mirror，并标记Primary为’d’，Mirror进入 ChangeTracking 状态。(详细的原理此处不作赘述，有兴趣可以参考本期月报的GPDB · 特性分析· GreenPlum Segment事务一致性与异常处理和上期的GPDB · 特性分析· GreenPlum FTS 机制)

当有Segment被标记为’d’后，Master将不会对其做处理，GP实例的启动（重启）也会将其忽略。这个时候，整个GP集群是处于有风险的状况中：

切过去的Mirror压力增大（需要做change tracking）；
节点单点，可靠性风险加大。

这个时候需要及时地对Segment进行修复。

GP的Segment修复

GP提供了一系列的控制脚本用于对GP进行操作，其中用于修复Segment的是gprecoverseg。使用方式比较简单，有限的几个主要参数如下：

-i

主要参数，用于指定一个配置文件，该配置文件描述了需要修复的Segment和修复后的目的位置。 - -F

可选项，指定后，gprecoverseg会将”-i”中指定的或标记”d”的实例删除，并从活着的Mirror复制一个完整一份到目标位置。 - -r

当FTS发现有Primary宕机并进行主备切换，在gprecoverseg修复后，担当Primary的Mirror角色并不会立即切换回来，就会导致部分主机上活跃的Segment过多从而引起性能瓶颈。因此需要恢复Segment原先的角色，称为re-balance。

举个使用的例子：

下面是一个正常的实例，


1. $ gpstate -s
2. /opt/python27/lib/python2.7/site-packages/Crypto/Util/number.py:57: PowmInsecureWarning: Not using mpz_powm_sec.  You should rebuild using libgmp >= 5 to avoid timing attack vulnerability.
3. _warn("Not using mpz_powm_sec.  You should rebuild using libgmp >= 5 to avoid timing attack vulnerability.", PowmInsecureWarning)
4. 20160418:21:39:29:016547 gpstate:host1:gpuser-[INFO]:-Starting gpstate with args: -s
5. 20160418:21:39:29:016547 gpstate:host1:gpuser-[INFO]:-local Greenplum Version: 'postgres (Greenplum Database) 4.3.99.00 build dev'
6. 20160418:21:39:29:016547 gpstate:host1:gpuser-[INFO]:-master Greenplum Version: 'PostgreSQL 8.3 (Greenplum Database 4.3.99.00 build dev) compiled on Apr 11 2016 22:02:39'
7. 20160418:21:39:29:016547 gpstate:host1:gpuser-[INFO]:-Obtaining Segment details from master...
8. 20160418:21:39:29:016547 gpstate:host1:gpuser-[INFO]:-Gathering data from segments...
9. .
10. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-----------------------------------------------------
11. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:--Master Configuration & Status
12. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-----------------------------------------------------
13. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master host                    = host1
14. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master postgres process ID     = 72447
15. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master data directory          = /workspace/gpuser/3007
16. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master port                    = 3007
17. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master current role            = dispatch
18. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Greenplum initsystem version   = 4.3.99.00 build dev
19. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Greenplum current version      = PostgreSQL 8.3 (Greenplum Database 4.3.99.00 build dev) compiled on Apr 11 2016 22:02:39
20. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Postgres version               = 8.3
21. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Master standby                 = host2
22. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Standby master state           = Standby host passive
23. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-----------------------------------------------------
24. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-Segment Instance Status Report
25. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-----------------------------------------------------
26. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Segment Info
27. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Hostname                          = host1
28. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Address                           = host1
29. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Datadir                           = /workspace/gpuser/3008
30. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Port                              = 3008
31. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Mirroring Info
32. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Current role                      = Primary
33. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Preferred role                    = Primary
34. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Mirror status                     = Synchronized
35. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Status
36. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      PID                               = 72388
37. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Configuration reports status as   = Up
38. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Database status                   = Up
39. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-----------------------------------------------------

41. ......

43. [INFO]:-----------------------------------------------------
44. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Segment Info
45. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Hostname                          = host1
46. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Address                           = host1
47. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Datadir                           = /workspace/gpuser/3012
48. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Port                              = 3012
49. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Mirroring Info
50. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Current role                      = Mirror
51. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Preferred role                    = Mirror
52. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Mirror status                     = Synchronized
53. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-   Status
54. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      PID                               = 75247
55. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Configuration reports status as   = Up
56. 20160418:21:39:30:016547 gpstate:host1:gpuser-[INFO]:-      Segment status                    = Up

选择一个kill之后（如3012这个端口的实例），执行gprecoverseg，如下：


1. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3008
2. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3008
3. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3014
4. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3014
5. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3010
6. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3010
7. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3015
8. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3015
9. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3008
10. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3008
11. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3011
12. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3011
13. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3013
14. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3013
15. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3012
16. 20160418:21:40:58:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3012

18. ......

20. 20160418:21:41:18:017989 gpstate:host1:gpuser-[DEBUG]:-[worker6] finished cmd: Get segment status cmdStr='sshpass -e ssh -o 'StrictHostKeyChecking no' host1 ". /workspace/gpdb/greenplum_path.sh; $GPHOME/bin/gp_primarymirror -h host1 -p 3012"'  had result: cmd had rc=15 completed=True halted=False
21. stdout=''
22. stderr='failed to connect: Connection refused (errno: 111)

24. Retrying no 1

26. failed to connect: Connection refused (errno: 111)

28. Retrying no 2

30. ......

32. 20160418:21:41:18:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Encountered error Not ready to connect to database mode: PrimarySegment

34. segmentState: Fault
35. dataState: InSync
36. faultType: FaultMirror
37. mode: PrimarySegment
38. segmentState: Fault
39. dataState: InSync
40. faultType: FaultMirror

这个时候连接这个实例去获取信息是失败的，失败的原因后面再说。这个时候失败后会重试5次，当再一次尝试的时候发现了不同：


1. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3008
2. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3008
3. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3014
4. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3014
5. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3010
6. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3010
7. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3015
8. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3015
9. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3008
10. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3008
11. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host1 -p 3011
12. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host1 -p 3011
13. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Sending msg getStatus and cmdStr $GPHOME/bin/gp_primarymirror -h host2 -p 3013
14. 20160418:21:41:23:017989 gprecoverseg:host1:gpuser-[DEBUG]:-Adding cmd to work_queue: $GPHOME/bin/gp_primarymirror -h host2 -p 3013

会发现，少了一个Segment的命令，而这个Segment正是刚才kill的Segment。继续往下看执行结果，gprecoverseg执行了下面的內容：


1. 20160418:23:16:20:085203 gprecoverseg:host1:gpuser-[DEBUG]:-[worker7] finished cmd: Get segment status information cmdStr='sshpass -e ssh -o 'StrictHostKeyChecking no' host2 ". /workspace/gpdb/greenplum_path.sh; $GPHOME/bin/gp_primarymirror -h host2 -p 3013"'  had result: cmd had rc=1 completed=True halted=False
2. stdout=''
3. stderr='mode: PrimarySegment
4. segmentState: Ready
5. dataState: InChangeTracking
6. faultType: NotInitialized
7. mode: PrimarySegment
8. segmentState: Ready
9. dataState: InChangeTracking
10. faultType: NotInitialized
11. '

这个实例为什么单独检查呢？而且这个时候如果失败，则会直接退出无法继续执行。

在一系列的检查之后，先更新catalog中的操作记录表：


1. UPDATE pg_catalog.gp_segment_configuration

之后即调用命令进行数据的恢复：


1. /workspace/gpdb/bin/lib/gpconfigurenewsegment -c /workspace/gpuser/3012:3012:false:false:9 -v -B 16 --write-gpid-file-only

最后再启动Segment，并更新catalog：


1. $GPHOME/sbin/gpsegstart.py -C en_US.utf8:C:C -M quiescent -V 'postgres (Greenplum Database) 4.3.99.00 build dev' -n 4 --era df86ca11ca2fc214_160418165251 -t 600 -v -p KGRwMApTJ2Ric0J5UG9ydCcKcDEKKGRwMgpJMzAxMgooZHAzClMndGFyZ2V0TW9kZScKcDQKUydtaXJyb3InCnA1CnNTJ2RiaWQnCnA2Ckk5CnNTJ2hvc3ROYW1lJwpwNwpTJzEwLjk3LjI0OC43MycKcDgKc1MncGVlclBvcnQnCnA5CkkzNTEzCnNTJ3BlZXJQTVBvcnQnCnAxMApJMzAxMwpzUydwZWVyTmFtZScKcDExClMncnQxYjA3MDI0LnRiYycKcDEyCnNTJ2Z1bGxSZXN5bmNGbGFnJwpwMTMKSTAwCnNTJ21vZGUnCnAxNApTJ3InCnAxNQpzUydob3N0UG9ydCcKcDE2CkkzNTEyCnNzcy4= -D '9|3|m|m|r|d|host1|host1|3012|3512|/workspace/gpuser/3012||'

3. ......

5. 20160419:01:21:05:042692 gprecoverseg:host1:gpuser-[DEBUG]:-UPDATE pg_catalog.gp_segment_configuration
6. SET
7. mode = 'r',
8. status = 'u'
9. WHERE dbid = 5
10. 20160419:01:21:05:042692 gprecoverseg:host1:gpuser-[DEBUG]:-INSERT INTO gp_configuration_history (time, dbid, "desc") VALUES(
11. now(),
12. 5,
13. 'gprecoverseg: segment resync marking mirrors up and primaries resync: segment mode and status'
14. )
15. 20160419:01:21:05:042692 gprecoverseg:host1:gpuser-[DEBUG]:-UPDATE pg_catalog.gp_segment_configuration
16. SET
17. mode = 'r',
18. status = 'u'
19. WHERE dbid = 9
20. 20160419:01:21:05:042692 gprecoverseg:host1:gpuser-[DEBUG]:-INSERT INTO gp_configuration_history (time, dbid, "desc") VALUES(
21. now(),
22. 9,
23. 'gprecoverseg: segment resync marking mirrors up and primaries resync: segment mode and status'
24. )
25. 20160419:01:21:05:042692 gprecoverseg:host1:gpuser-[DEBUG]:-UPDATE gp_fault_strategy

这样即是一个完整的gprecoverseg过程。执行过后，对应的Primary和Mirror会进入”r”状态，表示正在做数据同步。

下面来看其中的详细步骤和原理。

实现原理

上面的例子中，遗留了几个问题：

在gprecoverseg过程中，第一次获取Segment状态是不对的；
第二次获取Segment信息，比第一次少了一条；
单独检查了“-h host2 -p 3013”这个实例。

这几个问题在了解了原理后就很容易理解了。想要了解原理，可以先看下执行的步骤。从代码看来，其大致的步骤如下：

参数处理

GP的脚本用了较多的环境变量，且不同的脚本、不同的地方略有不同。如gprecoverseg用的就是MASTER_DATA_DIRECTORY，从MASTER_DATA_DIRECTORY指定的目录中得到Master相关的信息（如port）以进行相关操作。

gprecoverseg的参数，最重要的莫过于”-i”了，其指定了需要做修复的Segment，并且可以指定到不同的主机上，例如：


1. filespaceOrder=
2. host1:3012:/workspace/gpuser/3012 host2:3012:3512:/workspace/gpuser/3012

具体执行不再赘述。

判断Segment当前的状态

调用gp_primarymirror，向活着的segment发送消息，以判断Segment当前的状态。这是非常重要的一步，也是遇到问题最多的一步，经常会出现问题”Unable to connect to database”。事实上，造成这个失败的原因有很多，比较多的是：

其对应的Primary（Mirror）也宕机；
其对应的Primary的状态不对，如已经有gprecoverseg在进行（或执行失败，状态出问题等）。

在做这一步的时候，是依赖gp_segment_configuration中的数据的，即会首先从GP　Master上获取相应的数据，与下一步中的描述基本相同。

如果这个Segment被标记为”d”，那么是不会向该Segment发起状态信息请求。

而如果对应的Primary/Mirror都宕机了，他们的状态不会同时为”d”（有可能都为”u”，比如同时异常的时候，FTS不会更新他们）。因此对标记为”u”实际已经宕机的Segment连接获取状态信息的时候，则会报错。这个时候就不是gprecoverseg所能处理的问题了，只能重启整个实例。

回到前面的问题。第一次执行失败即因为Segment的状态尚未更新；第二次执行少了一个Segment，即状态被更新为”d”后不进行连接。

在检查完所有状态为”u”的Segment连接后，则会针对宕机的Mirror进行检查，查看其对应的主库是否正常，可以用于修复数据，即是第三个问题的答案。如：


1. stdout=''
2. stderr='mode: PrimarySegment
3. segmentState: Ready
4. dataState: InSync
5. faultType: NotInitialized
6. mode: PrimarySegment
7. segmentState: Ready
8. dataState: InSync
9. faultType: NotInitialized
10. '

或者这样


1. stdout=''
2. stderr='mode: PrimarySegment
3. segmentState: Ready
4. dataState: InChangeTracking
5. faultType: NotInitialized
6. mode: PrimarySegment
7. segmentState: Ready
8. dataState: InChangeTracking
9. faultType: NotInitialized
10. '

正常情况下，当Mirror出现问题，Primary发现后会进入ChangeTracking的状态。在这个状态里，Primary会记录下切换状态时间点之后的变更，用于当Mirror恢复时进行数据同步，而不用每次都做一次全量。

从master获取segment的信息

包括IP、PORT、ROLE、Status、数据目录、临时空间等，如下：


1. dbid | content | role | preferred\_role | mode | status |   hostname    |    address    | port | replication\_port | oid  |   fselocation
2. ------+---------+------+----------------+------+--------+---------------+---------------+------+------------------+------+-----------------
3. 1 |      -1 | p    | p              | s    | u      | host1 | host1  | 3007 |                  | 3052 | /workspace/gpuser/3007
4. 10 |      -1 | m    | m              | s    | u      | host2  | host2  | 3007 |                  | 3052 | /workspace/gpuser/3007
5. 2 |       0 | p    | p              | s    | u      | host1 | host1 | 3008 |             3508 | 3052 | /workspace/gpuser/3008
6. 6 |       0 | m    | m              | s    | u      | host2 | host2  | 3014 |             3514 | 3052 | /workspace/gpuser/3014
7. 3 |       1 | p    | p              | s    | u      | host1 | host1 | 3010 |             3510 | 3052 | /workspace/gpuser/3010
8. 7 |       1 | m    | m              | s    | u      | host2 | host2  | 3015 |             3515 | 3052 | /workspace/gpuser/3015
9. 4 |       2 | p    | p              | s    | u      | host2 | host2 | 3008 |             3508 | 3052 | /workspace/gpuser/3008
10. 8 |       2 | m    | m              | s    | u      | host1 | host1  | 3011 |             3511 | 3052 | /workspace/gpuser/3011
11. 5 |       3 | p    | p              | s    | u      | host2 | host2 | 3013 |             3513 | 3052 | /workspace/gpuser/3013
12. 9 |       3 | m    | m              | s    | u      | host1 | host1  | 3012 |             3512 | 3052 | /workspace/gpuser/3012

IP/PORT/ROLE/STATUS/目录/FILESPACE等信息，后面的Mirror修复列表、临时空间、操作对象的信息都依赖于此。

修复准备

在获取所有的Segment信息后，会针对配置文件、参数等相关信息确定，包括：

Segment修复对象

确定Segment修复对象和数据源，即Primary；需要修复的Segment有可能是多个。并获取需要修复的Segment的相关信息，包括端口、流复制端口、数据目录、临时空间、文件空间等信息，以及是否强制修复等。 - 主机环境

在获取所需要修复的Segment列表后，需要确保所在主机环境是可以的，包括端口占用、目录的占用等有可能冲突的地方。

如果没有指定主机，则会在已有的主机中选择一个。

修复

修复的步骤是：

关闭宕机的Mirror，并清理shared memory
确定需要修复的Segment已经被标记为”d”
如有需要，则进行删除，如”-F”的情况
打包压缩、复制数据到目标位置
关闭SIGINT、SIG_IGN，更新元数据库，打开SIGINT、SIG_IGN

以上步骤后，即可实现对Segment的本地（in-place）或跨机修复。

re-balance

当修复完Segment之后，原先因为Primary宕机而切到Mirror上的Segment并不会主动切回来，这个时候有可能出现性能倾斜而影响性能，因此需要做”re-balance”，执行：


1. gprecoverseg -r

执行该命令会将role切换为preferred_role，保证整个集群的角色平衡而不致于部分主机Primay更多引起性能瓶颈。

原文：http://mysql.taobao.org/monthly/2016/04/03/