南大通用数据库-Gbase-8a-报错集锦-01-Can‘t connect to GBase server或Cannot assign requested address

目录

 一、测试版本

二、实验模拟

1、检查操作系统参数

2、gccli执行SQL

3、查看socket 统计信息

三、解决方法

1、修改/etc/sysctl.conf

2、内核参数生效

---

 一、测试版本

名称	值
CPU	Intel(R) Core(TM) i5-1035G1 CPU @ 1.00GHz
操作系统	CentOS Linux release 7.9.2009 (Core)
内存	3G
逻辑核数	2
Gbase8a版本	8.6.2-R43

二、实验模拟

1、检查操作系统参数

[root@czg2 ~]# cat /proc/sys/net/ipv4/ip_local_port_range
60996   61000
[root@czg2 ~]# cat /proc/sys/net/ipv4/tcp_fin_timeout
60

参数名	描述
net.ipv4.ip_local_port_range	决定了客户端的一个 ip 可用的端口数量。
net.ipv4.tcp_fin_timeout	如果套接字由本端要求关闭，这个参数决定了它保持在FIN-WAIT-2状态的时间。意思就是客户端关闭了，需要保持一个等待状态一段时间，才释放。

net.ipv4.ip_local_port_range这个参数是为了模拟现象，而故意配置的这么小，60996到61000一共五个可用的端口数量。

2、gccli执行SQL

这里我们执行六次，复现报错。

[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
+------+-----------------+-------+
| a    | CURRENCY_TYPE_5 | moon2 |
+------+-----------------+-------+
| 光   | B               |  NULL |
+------+-----------------+-------+
[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
+------+-----------------+-------+
| a    | CURRENCY_TYPE_5 | moon2 |
+------+-----------------+-------+
| 光   | B               |  NULL |
+------+-----------------+-------+
[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
+------+-----------------+-------+
| a    | CURRENCY_TYPE_5 | moon2 |
+------+-----------------+-------+
| 光   | B               |  NULL |
+------+-----------------+-------+
[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
+------+-----------------+-------+
| a    | CURRENCY_TYPE_5 | moon2 |
+------+-----------------+-------+
| 光   | B               |  NULL |
+------+-----------------+-------+
[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
+------+-----------------+-------+
| a    | CURRENCY_TYPE_5 | moon2 |
+------+-----------------+-------+
| 光   | B               |  NULL |
+------+-----------------+-------+
[gbase@czg2 ~]$ gccli -e 'select * from czg.sun limit 10;'
ERROR 1708 (HY000) at line 1: [192.168.142.12:5050](GBA-02AD-0005)Failed to query in gnode:
DETAIL: Can't connect to GBase server on '::ffff:192.168.142.12' (99)

3、查看socket 统计信息

[root@czg2 ~]# ss -an |grep :5050
tcp    LISTEN     0      32767  [::]:5050               [::]:*                  
tcp    TIME-WAIT  0      0       [::ffff:192.168.142.12]:60999               [::ffff:192.168.142.12]:5050               
tcp    TIME-WAIT  0      0       [::ffff:192.168.142.12]:60996               [::ffff:192.168.142.12]:5050               
tcp    TIME-WAIT  0      0       [::ffff:192.168.142.12]:60998               [::ffff:192.168.142.12]:5050               
tcp    TIME-WAIT  0      0       [::ffff:192.168.142.12]:60997               [::ffff:192.168.142.12]:5050               
tcp    TIME-WAIT  0      0       [::ffff:192.168.142.12]:61000               [::ffff:192.168.142.12]:5050    

LISTEN：被动打开。

ESTAB：数据发送状态。

TIME-WAIT：等待。

上面的三个状态解释，如果有不对的，请多纠正。

确实可以看到五个等待，那我们在排查这个问题时，就可以按照这个思路去排查。查看内核参数net.ipv4.ip_local_port_range再用ss，看是否差不多，如果差不多，可以初步判断是应用端在频繁的连服务器，并且每次连接后，都又在较短的时间里断开，导致出现过多的TIME_WAIT，消耗完了可用的端口号，以至于新的连接没办法绑定端口。

三、解决方法

我测试的版本报的是Can't connect to GBase server。有的版本报的是Cannot assign requested address。问题的原因都有可能是上面分析的结果。

1、修改/etc/sysctl.conf

修改参数net.ipv4.ip_local_port_range = 30000    61000，net.ipv4.tcp_fin_timeout = 30

[root@czg2 gcinstall]# cat /etc/sysctl.conf
kernel.core_uses_pid = 1
net.core.netdev_max_backlog = 262144
net.core.rmem_default = 8388608
net.core.rmem_max = 16777216
net.core.wmem_default = 8388608
net.core.wmem_max = 16777216
net.ipv4.tcp_max_syn_backlog = 262144
net.ipv4.tcp_rmem = 4096 87380 4194304
net.ipv4.tcp_sack = 1
net.ipv4.ip_local_reserved_ports = 5050,5258,5288,6666,6268
net.ipv4.tcp_syncookies = 1
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_wmem = 4096 16384 4194304
vm.vfs_cache_pressure = 1024
vm.swappiness = 1
vm.overcommit_memory = 0
vm.zone_reclaim_mode = 0
vm.max_map_count = 188529
net.core.somaxconn = 32767
net.ipv4.tcp_fin_timeout = 30
net.ipv4.tcp_max_orphans = 3276800
net.ipv4.tcp_max_tw_buckets = 20000
net.ipv4.tcp_mem = 94500000 915000000 927000000
net.ipv4.tcp_timestamps = 0
net.ipv4.tcp_tw_recycle = 1
net.ipv4.tcp_tw_reuse = 1
#vm.min_free_kbytes = 4096 #Commented out by gcluster
vm.min_free_kbytes = 301645
net.ipv4.ip_local_port_range = 30000    61000

2、内核参数生效

[root@czg2 gcinstall]# sysctl -p
kernel.core_uses_pid = 1
net.core.netdev_max_backlog = 262144
net.core.rmem_default = 8388608
net.core.rmem_max = 16777216
net.core.wmem_default = 8388608
net.core.wmem_max = 16777216
net.ipv4.tcp_max_syn_backlog = 262144
net.ipv4.tcp_rmem = 4096 87380 4194304
net.ipv4.tcp_sack = 1
net.ipv4.ip_local_reserved_ports = 5050,5258,5288,6666,6268
net.ipv4.tcp_syncookies = 1
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_wmem = 4096 16384 4194304
vm.vfs_cache_pressure = 1024
vm.swappiness = 1
vm.overcommit_memory = 0
vm.zone_reclaim_mode = 0
vm.max_map_count = 188529
net.core.somaxconn = 32767
net.ipv4.tcp_fin_timeout = 30
net.ipv4.tcp_max_orphans = 3276800
net.ipv4.tcp_max_tw_buckets = 20000
net.ipv4.tcp_mem = 94500000 915000000 927000000
net.ipv4.tcp_timestamps = 0
net.ipv4.tcp_tw_recycle = 1
net.ipv4.tcp_tw_reuse = 1
vm.min_free_kbytes = 301645
net.ipv4.ip_local_port_range = 30000    61000