# 检查Segment状态
$ gpstate -e

# 输出日志：
20260408:10:00:00:gpstate:mdw:fgedu-[INFO]:-GreenPlum instance status
20260408:10:00:05:gpstate:mdw:fgedu-[INFO]:-Segment sdw1:40000 is down

# 查看故障详情
$ gpstate -m

# 输出日志：
20260408:10:01:00:gpstate:mdw:fgedu-[INFO]:-Mirror Segment status
20260408:10:01:05:gpstate:mdw:fgedu-[INFO]:-sdw1:40000 is down
20260408:10:01:10:gpstate:mdw:fgedu-[INFO]:-sdw2:50000 is acting as primary

# 恢复Segment
$ gprecoverseg

# 输出日志：
20260408:10:05:00:gprecoverseg:mdw:fgedu-[INFO]:-Starting gprecoverseg
20260408:10:05:05:gprecoverseg:mdw:fgedu-[INFO]:-Recovering segment: sdw1:40000
20260408:10:30:00:gprecoverseg:mdw:fgedu-[INFO]:-Segment recovery completed

# 验证恢复结果
$ gpstate -e

# 输出日志：
20260408:10:31:00:gpstate:mdw:fgedu-[INFO]:-All segments are synchronized

学习交流加群风哥QQ113257174

# 查看Master日志
$ tail -100 $MASTER_DATA_DIRECTORY/log/gpdb-*.csv

# 输出日志：
2026-04-08 10:00:00.000 CST,,,p12345,th-1,,,,0 LOG: database system is ready to accept connections
2026-04-08 10:05:00.000 CST,,,p12346,th-1,,,,0 ERROR: relation “fgedu_sales” does not exist

# 查看Segment日志
$ ssh sdw1 “tail -100 /GreenPlum/fgdata/log/gpdb-*.csv”

# 输出日志：
2026-04-08 10:00:00.000 CST,,,p12345,th-1,,,,0 LOG: segment is ready
2026-04-08 10:05:00.000 CST,,,p12346,th-1,,,,0 ERROR: could not write to file

# 分析错误日志
$ grep -i “error\|fatal\|panic” $MASTER_DATA_DIRECTORY/log/gpdb-*.csv | tail -20

# 输出日志：
2026-04-08 10:05:00.000 CST,,,p12346,th-1,,,,0 ERROR: relation “fgedu_sales” does not exist
2026-04-08 10:10:00.000 CST,,,p12347,th-1,,,,0 FATAL: connection limit exceeded

# 查看当前活动会话
$ psql -d fgedudb -c “SELECT * FROM pg_stat_activity WHERE state = ‘active’;”

# 输出日志：
datid | datname | pid | usename | application_name | client_addr | state | query
——-+———-+——+———+——————+————-+——–+——-
16384 | fgedudb | 12345| fgedu | psql | 192.168.1.1| active | SELECT * FROM fgedu_sales

更多学习教程公众号风哥教程itpux_com

# 查询超时故障案例

# 1. 故障现象
# – 查询执行时间过长
# – 系统响应缓慢
# – 用户投诉

# 2. 故障排查
# 查看当前运行的查询
psql -d fgedudb -c “SELECT pid, now() – pg_stat_activity.query_start AS duration, query FROM pg_stat_activity WHERE state = ‘active’;”

# 输出日志：
pid | duration | query
——+—————–+———————————————————
12345| 00:30:00.000000 | SELECT * FROM fgedu_sales WHERE customer_id = 1001

# 3. 分析执行计划
psql -d fgedudb -c “EXPLAIN ANALYZE SELECT * FROM fgedu_sales WHERE customer_id = 1001;”

# 输出日志：
QUERY PLAN
————————————————————————————————–
Gather Motion 2:1 (slice1; segments: 2) (cost=0.00..1000000.00 rows=1000 width=100) (actual time=0.000..1800000.000 rows=1 loops=1)
-> Seq Scan on fgedu_sales (cost=0.00..1000000.00 rows=1000 width=100) (actual time=0.000..1800000.000 rows=1 loops=1)
Filter: (customer_id = 1001)
Rows Removed by Filter: 10000000

# 4. 解决方案
# 创建索引
CREATE INDEX idx_customer_id ON fgedu.fgedu_sales (customer_id);

# 更新统计信息
ANALYZE fgedu.fgedu_sales;

# 5. 验证效果
# 查询时间从30分钟降低到1秒

from GreenPlum视频:www.itpux.com

故障排查技巧：

1. 信息收集
– 收集错误信息
– 分析日志文件
– 查看系统状态
– 了解操作历史

2. 问题定位
– 使用gpstate检查状态
– 使用gp_toolkit分析
– 查看系统日志
– 分析执行计划

3. 解决方案
– 风哥教程参考官方文档
– 查找已知问题
– 制定解决方案
– 验证解决效果

4. 预防措施
– 完善监控告警
– 定期巡检维护
– 建立知识库
– 培训相关人员

5. 最佳实践
– 建立故障处理流程
– 保留故障日志
– 总结故障经验
– 持续改进优化

本文档介绍了GreenPlum故障排查与处理的核心内容，包括常见故障处理、日志分析、故障排查案例等，希望对大家有所帮助。