案例说明：
在主备库建立nfs共享存储的文件系统，作为sys_rman备份的repo-path，在备库作为repo-path节点执行备份，出现数据库连接到'5432端口的错误'，数据库实际的服务端口为54321。
适用版本：
KingbaseES V8R6
节点信息：
[kingbase@node102 bin]$ cat /etc/hosts
192.168.1.101 node101 # primary
192.168.1.102 node102 # standby
192.168.1.103 node103 # nfs server
一、配置NFS共享server
[root@node103 ~]# service nfs start
Redirecting to /bin/systemctl start nfs.service
# 查看共享存储
[root@node103 ~]# cat /etc/exports
/shdata 192.168.1.101(rw,no_root_squash) 192.168.1.102(rw,no_root_squash)
[root@node103 ~]# exportfs -v
/shdata 192.168.1.101(rw,wdelay,no_root_squash,no_subtree_check,sec=sys,rw,secure,no_root_squash,no_all_squash)
/shdata 192.168.1.102(rw,wdelay,no_root_squash,no_subtree_check,sec=sys,rw,secure,no_root_squash,no_all_squash)
[root@node103 ~]# df -h
Filesystem Size Used Avail Use% Mounted on
......
/dev/mapper/myvg-mylv 20G 223M 20G 2% /shdata
# 共享文件系统属主和权限
[root@node103 ~]# id kingbase
uid=200(kingbase) gid=1001(kingbase) groups=1001(kingbase)
[root@node103 ~]# ls -lhd /shdata
drwxr-x--- 7 kingbase kingbase 4.0K Jan 12 17:43 /shdata
二、nfs client挂载共享文件系统
1、主库配置
[root@node101 ~]# showmount -e 192.168.1.103
Export list for 192.168.1.103:
/shdata 192.168.1.102,192.168.1.101
[root@node101 ~]# mkdir -p /data/kingbase/bk
[root@node101 ~]# chown kingbase.kingbase /data/kingbase/bk
[root@node101 ~]# mount 192.168.1.103:/shdata /data/kingbase/bk
[root@node101 ~]# ls -lhd /data/kingbase/bk
drwxr-x--- 2 kingbase kingbase 6 Apr 13 11:28 /data/kingbase/bk
2、备库配置
[root@node102 ~]# showmount -e 192.168.1.103
Export list for 192.168.1.103:
/shdata 192.168.1.102,192.168.1.101
[root@node102 ~]# mkdir -p /data/kingbase/bk
[root@node102 ~]# mount 192.168.1.103:/shdata /data/kingbase/bk
[root@node102 ~]# ls -lhd /data/kingbase/bk
drwxr-x--- 2 kingbase kingbase 6 Apr 13 11:28 /data/kingbase/bk
[root@node102 ~]# ls -lh /data/kingbase/bk
total 0
三、集群节点信息
[kingbase@node102 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | LSN_Lag | Connection string
----+-------+---------+-----------+----------+----------+----------+----------+---------+---------------------------------------------------------------------------------------------------------------------------------------------------
1 | node1 | standby | running | node2 | default | 100 | 4 | 0 bytes | host=192.168.1.101 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node2 | primary | * running | | default | 100 | 5 | | host=192.168.1.102 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
四、配置sys_backup.conf
在备库执行备份，并将备份存储到nfs共享的文件系统下：
[kingbase@node102 bin]$ cat sys_backup.conf |grep -v ^$|grep -v ^#
_target_db_style="cluster"
_one_db_ip="192.168.1.101"
_repo_ip="192.168.1.102"
_stanza_name="kingbase"
_os_user_name="kingbase"
_repo_path="/data/kingbase/bk/c7_bk"
_repo_retention_full_count=5
_crond_full_days=7
_crond_diff_days=0
_crond_incr_days=1
_crond_full_hour=2
_crond_diff_hour=3
_crond_incr_hour=4
_band_width=0
_os_ip_cmd="/sbin/ip"
_os_rm_cmd="/bin/rm"
_os_sed_cmd="/bin/sed"
_os_grep_cmd="/bin/grep"
_single_data_dir="/home/kingbase/cluster/R6HA/ha7/kingbase/data"
_single_bin_dir="/home/kingbase/cluster/R6HA/ha7/kingbase/kingbase/bin"
_single_db_user="system"
_single_db_port="54321"
_use_scmd=off
_start_fast=y
_compress_type=none
_non_archived_space=1024
五、执行sys_backup.sh init
[kingbase@node102 bin]$ ./sys_backup.sh init
# pre-condition: check the non-archived WAL files
# generate local sys_rman.conf...DONE
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...DONE
# create stanza and check...(maybe 60+ seconds)
ERROR: create stanza failed, check log file /home/kingbase/cluster/R6HA/ha7/kingbase/kingbase/log/sys_rman_stanza-create.log
备份故障日志：
[kingbase@node102 bin]$ cat /home/kingbase/cluster/R6HA/ha7/kingbase/kingbase/log/sys_rman_stanza-create.log
2023-04-13 11:41:50.915 P00 INFO: stanza-create command begin 2.27: --band-width=0 --config=/data/kingbase/bk/c7_bk/sys_rman.conf --exec-id=27638-be67f3b4 --log-level-console=info --log-level-file=info --log-path=/home/kingbase/cluster/R6HA/ha7/kingbase/kingbase/log --log-subprocess --kb1-path=/home/kingbase/cluster/R6HA/ha7/kingbase/data --repo1-host=192.168.1.102 --repo1-host-config=/data/kingbase/bk/c7_bk/sys_rman.conf --repo1-host-user=kingbase --repo1-path=/data/kingbase/bk/c7_bk --stanza=kingbase
WARN: unable to check kb-1: [DbConnectError] unable to connect to 'dbname='test' port=5432': could not connect to server: No such file or directory
Is the server running locally and accepting
connections on Unix domain socket "/tmp/.s.KINGBASE.5432"?
ERROR: [056]: unable to find primary cluster - cannot proceed
2023-04-13 11:41:50.915 P00 INFO: stanza-create command end: aborted with exception [056]
六、解决方案
方案1：
Tips：
由于主备非repo节点，不存储备份，只存储sys_rman.conf文件，不占用磁盘空间，可以考虑在本地节点建立存储目录。
1、umount主库的nfs共享文件系统
[root@node101 ~]# umount /data/kingbase/bk
[root@node101 ~]# ls -lhd /data/kingbase/bk
drwxr-xr-x 3 kingbase kingbase 17 Mar 29 11:59 /data/kingbase/bk
2、执行备份初始化
[kingbase@node102 bin]$ ./sys_backup.sh init
# pre-condition: check the non-archived WAL files
# generate local sys_rman.conf...DONE
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...DONE
# create stanza and check...(maybe 60+ seconds)
# create stanza and check...DONE
# initial first full backup...(maybe several minutes)
# initial first full backup...DONE
# Initial sys_rman OK.
'sys_backup.sh start' should be executed when need back-rest feature.
'sys_backup.sh start' will add CRONTAB items.
Or you can manual backup once with user-guide.
方案2：
1）nfs server共享文件系统创建子目录
[kingbase@node103 ~]$ mkdir -p /shdata/bk{1,2}
[kingbase@node103 ~]$ ls -lh /shdata
total 0
drwxrwxr-x 2 kingbase kingbase 6 Apr 13 16:12 bk1
drwxrwxr-x 2 kingbase kingbase 6 Apr 13 16:12 bk2
2）nfs client挂载共享子目录
主库：
[root@node101 ~]# mount 192.168.1.103:/shdata/bk1 /data/kingbase/bk
[root@node101 ~]# df -h
.......
192.168.1.103:/shdata/bk1 20G 193M 20G 1% /data/kingbase/bk
备库：
[root@node102 ~]# mount 192.168.1.103:/shdata/bk2 /data/kingbase/bk
[root@node102 ~]# df -h
Filesystem Size Used Avail Use% Mounted on
......
192.168.1.103:/shdata/bk2 20G 32M 20G 1% /data/kingbase/bk
3）备库执行sys_backup.sh init
[kingbase@node102 bin]$ ./sys_backup.sh init
# pre-condition: check the non-archived WAL files
# generate local sys_rman.conf...DONE
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...
# update all node: sys_rman.conf and archive_command with sys_rman.archive-push...DONE
# create stanza and check...(maybe 60+ seconds)
# create stanza and check...DONE
# initial first full backup...(maybe several minutes)
# initial first full backup...DONE
# Initial sys_rman OK.
'sys_backup.sh start' should be executed when need back-rest feature.
'sys_backup.sh start' will add CRONTAB items.
Or you can manual backup once with user-guide.
You have mail in /var/spool/mail/kingbase
4）查看备份信息
Tips：
备库作为repo，存储了所有备份，主库只是存储了sys_rman.conf。
备库：
[root@node102 ~]# ls -lh /data/kingbase/bk
total 0
drwxrwxr-x 4 kingbase kingbase 53 Apr 13 16:15 c7_bk
[root@node102 ~]# ls -lh /data/kingbase/bk/c7_bk/
total 4.0K
drwxr-x--- 3 kingbase kingbase 21 Apr 13 16:15 archive
drwxr-x--- 3 kingbase kingbase 21 Apr 13 16:15 backup
-rw-rw-r-- 1 kingbase kingbase 850 Apr 13 16:15 sys_rman.conf
主库：
[root@node101 ~]# ls -lh /data/kingbase/bk
total 0
drwxrwxr-x 2 kingbase kingbase 26 Apr 13 16:15 c7_bk
[root@node101 ~]# ls -lh /data/kingbase/bk/c7_bk
total 4.0K
-rw-rw-r-- 1 kingbase kingbase 451 Apr 13 16:15 sys_rman.conf
七、总结
在集群环境下，如果备库作为repo-path以‘cluster’模式执行备份，将同时会在主备库都建立repo-path对应的目录，在主库的repo-path中写入sys_rman.conf，在备库的repo-path中存储备份和写入sys_rman.conf；
如果主备库使用是nfs共享存储，将在写入数据时出现问题，对nfs共享存储备份的支持还需要进一步支持和细化。