KingbaseES V8R3 集群运维案例之 -- 主库归档失败
案例说明：
本案例用于KingbaseES V8R3集群归档进程归档日志失败的处理，对于一线的生产环境具有 一定的参考意义。
适用版本：
KingbaseES V8R3
集群架构：
一、案例故障现象（主库归档失败）
1、主库数据库服务进程
[kingbase@node1 bin]$ ps -ef |grep kingbase
kingbase 8180 1 0 13:44 ? 00:00:00 /home/kingbase/cluster/kha/db/bin/kingbase -D /home/kingbase/cluster/kha/db/data
kingbase 8181 8180 0 13:44 ? 00:00:00 kingbase: logger process
kingbase 8183 8180 0 13:44 ? 00:00:00 kingbase: checkpointer process
kingbase 8184 8180 0 13:44 ? 00:00:00 kingbase: writer process
kingbase 8185 8180 0 13:44 ? 00:00:00 kingbase: wal writer process
kingbase 8186 8180 0 13:44 ? 00:00:00 kingbase: autovacuum launcher process
kingbase 8187 8180 0 13:44 ? 00:00:00 kingbase: archiver process failed on 000000020000000000000010
.......
2、查看sys_log日志
2021-03-01 13:49:30.693 CST,,,8187,,603c7f31.1ffb,23,,2021-03-01 13:44:17 CST,,0,LOG,00000,"archive command failed with exit code 45","The failed archive command was: /home/kingbase/cluster/kha/db/bin/sys_rman_v6 --config /home/kingbase/kbbr3_repo/sys_rman_v6.conf --stanza=kingbase archive-push sys_xlog/000000020000000000000010",,,,,,,,""
2021-03-01 13:49:30.694 CST,,,8187,,603c7f31.1ffb,24,,2021-03-01 13:44:17 CST,,0,WARNING,01000,"archiving transaction log file ""000000020000000000000010"" failed too many times, will try again later",,,,,,,,,""
二、故障处理步骤
Tips：
在KingbaseES V8R3后期的版本sys_rman升级为sys_rman_v6,归档方式已经由早期的cp改为通过sys_rman的archive-push执行，归档文件将存储到sys_rman的备份目录下（repo-path）。
1、查看archive日志归档配置
2、查看备份配置文件和目录信息
[kingbase@node1 sys_log]$ ls -lh /home/kingbase/kbbr3_repo/sys_rman_v6.conf
-rw-rw-r-- 1 kingbase kingbase 589 Mar 1 12:26 /home/kingbase/kbbr3_repo/sys_rman_v6.conf
[kingbase@node1 sys_log]$ cat /home/kingbase/kbbr3_repo/sys_rman_v6.conf
# Genarate by script at 20210301122559, should not change manually
[kingbase]
.......
[global]
repo1-path=/home/kingbase/kbbr3_repo #数据文件备份及归档日志将存储到此目录下
repo1-retention-full=5
log-path=/tmp/
log-level-file=info
log-level-console=info
log-subprocess=y
process-max=4
#### default gz, support: gz none
compress-type=gz
compress-level=3
3、执行手工归档
[kingbase@node1 bin]$ /home/kingbase/cluster/kha/db/bin/sys_rman_v6 --config /home/kingbase/kbbr3_repo/sys_rman_v6.conf --stanza=kingbase archive-push /home/kingbase/cluster/kha/db/data/sys_xlog/000000020000000000000010
2021-03-01 14:43:20.928 P00 INFO: archive-push command begin 2.27: [/home/kingbase/cluster/kha/db/data/sys_xlog/000000020000000000000010] --compress-level=3 --compress-type=gz --config=/home/kingbase/kbbr3_repo/sys_rman_v6.conf --log-level-console=info --log-level-file=info --log-path=/tmp --log-subprocess --kb2-host=192.168.7.243 --kb1-path=/home/kingbase/cluster/kha/db/data --kb2-path=/home/kingbase/cluster/kha/db/data --process-max=4 --repo1-path=/home/kingbase/kbbr3_repo --stanza=kingbase
2021-03-01 14:43:21.203 P00 INFO: pushed WAL file '000000020000000000000010' to the archive
2021-03-01 14:43:21.204 P00 INFO: archive-push command end: completed successfully (276ms)
# 查看归档日志文件
[kingbase@node1 data]$ ls -lh /home/kingbase/kbbr3_repo/archive/kingbase/9.6-1/0000000200000000
total 208K
-rw-r----- 1 kingbase kingbase 91K Mar 1 12:20 00000002000000000000000E-583ac46b5270f365463cb0bfb3b96185af6492dd.gz
-rw-r----- 1 kingbase kingbase 303 Mar 1 12:20 00000002000000000000000F.00000028.backup
-rw-r----- 1 kingbase kingbase 83K Mar 1 12:20 00000002000000000000000F-1c450ca422ee6312e8a69dcb7d8c446a99425995.gz
-rw-r----- 1 kingbase kingbase 28K Mar 1 12:21 000000020000000000000010-044ab3927144a6510a42ce9c2bc331cf209aff56.gz
=从以上信息可知，手工归档成功，可以说明归档的配置文件及目录权限等没有问题。=
4、重启集群测试
[kingbase@node1 bin]$ ps -ef |grep kingbase
kingbase 8180 1 0 13:44 ? 00:00:00 /home/kingbase/cluster/kha/db/bin/kingbase -D /home/kingbase/cluster/kha/db/data
kingbase 8181 8180 0 13:44 ? 00:00:00 kingbase: logger process
kingbase 8183 8180 0 13:44 ? 00:00:00 kingbase: checkpointer process
kingbase 8184 8180 0 13:44 ? 00:00:00 kingbase: writer process
kingbase 8185 8180 0 13:44 ? 00:00:00 kingbase: wal writer process
kingbase 8186 8180 0 13:44 ? 00:00:00 kingbase: autovacuum launcher process
kingbase 8187 8180 0 13:44 ? 00:00:00 kingbase: archiver process failed on 000000020000000000000010
=== 从以上信息可知，归档仍然失败。===
5、修改archive_command配置（跳过归档）
6、重启集群测试
[kingbase@node1 bin]$ ps -ef |grep kingbase
kingbase 21906 5688 0 14:09 pts/0 00:00:00 ./ksql -U SYSTEM -W ******** TEST
kingbase 23167 1 0 14:11 ? 00:00:00 /home/kingbase/cluster/kha/db/bin/kingbase -D /home/kingbase/cluster/kha/db/data
kingbase 23168 23167 0 14:11 ? 00:00:00 kingbase: logger process
kingbase 23170 23167 0 14:11 ? 00:00:00 kingbase: checkpointer process
kingbase 23171 23167 0 14:11 ? 00:00:00 kingbase: writer process
kingbase 23172 23167 0 14:11 ? 00:00:00 kingbase: wal writer process
kingbase 23173 23167 0 14:11 ? 00:00:00 kingbase: autovacuum launcher process
kingbase 23174 23167 0 14:11 ? 00:00:00 kingbase: archiver process
=== 从以上信息可知，已经没有归档失败的状态信息。===
7、在主库手工执行wal日志切换
TEST=# select sys_switch_xlog();
SYS_SWITCH_XLOG
-----------------
0/180000E8
(1 row)
TEST=# select sys_switch_xlog();
SYS_SWITCH_XLOG
-----------------
0/19000078
(1 row)
TEST=# select sys_switch_xlog();
SYS_SWITCH_XLOG
-----------------
0/1A000000
(1 row)
8、再重新恢复archive_command配置
9、重启集群测试
[kingbase@node1 bin]$ ps -ef |grep kingbase
kingbase 25979 1 0 14:15 ? 00:00:00 /home/kingbase/cluster/kha/db/bin/kingbase -D /home/kingbase/cluster/kha/db/data
kingbase 25980 25979 0 14:15 ? 00:00:00 kingbase: logger process
kingbase 25983 25979 0 14:15 ? 00:00:00 kingbase: checkpointer process
kingbase 25984 25979 0 14:15 ? 00:00:00 kingbase: writer process
kingbase 25985 25979 0 14:15 ? 00:00:00 kingbase: wal writer process
kingbase 25986 25979 0 14:15 ? 00:00:00 kingbase: autovacuum launcher process
kingbase 25987 25979 0 14:15 ? 00:00:00 kingbase: archiver process
=== 从以上信息可知，已经没有归档失败的状态信息。===
10、测试wal日志归档
1）执行wal日志切换（主库）
TEST=# select sys_switch_xlog();
SYS_SWITCH_XLOG
-----------------
0/1C000078
(1 row)
TEST=# select sys_switch_xlog();
SYS_SWITCH_XLOG
-----------------
0/1E000000
(1 row)
2）查看归档信息
[kingbase@node1 0000000200000000]$ pwd
/home/kingbase/kbbr3_repo/archive/kingbase/9.6-1/0000000200000000
[kingbase@node1 0000000200000000]$ ls -lh
total 460K
-rw-r----- 1 kingbase kingbase 91K Mar 1 12:20 00000002000000000000000E-583ac46b5270f365463cb0bfb3b96185af6492dd.gz
-rw-r----- 1 kingbase kingbase 303 Mar 1 12:20 00000002000000000000000F.00000028.backup
-rw-r----- 1 kingbase kingbase 83K Mar 1 12:20 00000002000000000000000F-1c450ca422ee6312e8a69dcb7d8c446a99425995.gz
-rw-r----- 1 kingbase kingbase 28K Mar 1 12:21 000000020000000000000010-044ab3927144a6510a42ce9c2bc331cf209aff56.gz
-rw-r----- 1 kingbase kingbase 83K Mar 1 14:22 00000002000000000000001B-f64eadf9b3ecb50ce6925cdc8c196bf33af4cc8c.gz
-rw-r----- 1 kingbase kingbase 84K Mar 1 14:22 00000002000000000000001C-300ed282be050b1194cc15b974b5b11b120d2076.gz
-rw-r----- 1 kingbase kingbase 84K Mar 1 14:23 00000002000000000000001D-249ab2fafe02a51ca06846098d4bad2f786d1422.gz
=从归档日志信息看，在wal日志发生切换时，产生了归档；但是前面在修改archive_command='/bin/true'参数后，导致一部分wal日志没有归档。==
11、查看数据库服务进程
[kingbase@node1 0000000200000000]$ ps -ef |grep kingbase
.......
kingbase 25979 1 0 14:15 ? 00:00:00 /home/kingbase/cluster/kha/db/bin/kingbase -D /home/kingbase/cluster/kha/db/data
kingbase 25980 25979 0 14:15 ? 00:00:00 kingbase: logger process
kingbase 25983 25979 0 14:15 ? 00:00:00 kingbase: checkpointer process
kingbase 25984 25979 0 14:15 ? 00:00:00 kingbase: writer process
kingbase 25985 25979 0 14:15 ? 00:00:00 kingbase: wal writer process
kingbase 25986 25979 0 14:15 ? 00:00:00 kingbase: autovacuum launcher process
kingbase 25987 25979 0 14:15 ? 00:00:00 kingbase: archiver process last was 00000002000000000000001D
=== 从以上信息获知，数据库归档进程正常。===
三、总结
此案例解决了KingbaseES R3集群归档失败的故障，对于数据库归档失败在原生PostgreSQL也发现过此类问题，处理起来比较繁琐，主库归档恢复正常后，需要对数据库做一次物理的全备，因为中间缺失了一部分wal日志的归档。
KINGBASE研究院
