【磐维数据库】扩容报错：invalid literal for int() with base 10 '128'

问题

• 扩容报错：

[root@xxx script]# ./gs_expansion -U omm -G dbgrp -X /database/panweidb/soft/cluster_config_expansion.xml -h xxx
Start expansion with cluster manager component.
Traceback (most recent call last):
  File "/database/panweidb/soft/PanWeiDB/script/gs_expansion", line 591, in <module>
    expansion.expand_run()
  File "/database/panweidb/soft/PanWeiDB/script/gs_expansion", line 424, in expand_run
    expand_impl.run()
  File "/database/panweidb/soft/PanWeiDB/script/impl/expansion/expansion_impl_with_cm.py", line 552, in run
    self.expansion_check()
  File "/database/panweidb/soft/PanWeiDB/script/impl/expansion/expansion_impl_with_cm.py", line 451, in expansion_check
    self.checkClusterStatus()
  File "/database/panweidb/soft/PanWeiDB/script/impl/expansion/ExpansionImpl.py", line 1463, in checkClusterStatus
    self.walKeepSegments = int(currentWalKeepSegments)
ValueError: invalid literal for int() with base 10: "'128'"
复制

分析

首先分析 Traceback 信息：

• 先看最后一行，错误名称为 ValueError，错误内容为

invalid literal for int() with base 10: "'128'"
复制

说的是进行类型转换的 int() 函数，接收到了错误的参数 ‘128’；

• 从下往上看，最后执行的代码是

self.walKeepSegments = int(currentWalKeepSegments)
复制

说明获取到的 currentWalKeepSegments 是 ‘128’；

• 查看 postgresql.conf 中的该参数，有多余引号：

[omm@xxx ~]$ cat $PGDATA/postgresql.conf | grep wal_keep_segments
wal_keep_segments = '128'		# in logfile segments, 16MB each normal, 1GB each in share storage mode; 0 disables
复制

解决

修改该参数，去掉引号，reload 后生效。

其他

• 也可以继续往上看 Traceback，出错时，程序应该是处于扩容时的检查阶段，正在检查集群状态。

参考

Understanding the Python Traceback