From 109a9fcc960c5ba3dc091266c4c5d147e80607ab Mon Sep 17 00:00:00 2001
From: Lohit Kolluri <lohitkolluri@gmail.com>
Date: Sun, 10 Mar 2024 11:30:28 +0530
Subject: [PATCH] Translate 10-pg_clean.md

---
 docs/guide/10-pg_clean.md | 131 ++++++++++++++++++--------------------
 1 file changed, 61 insertions(+), 70 deletions(-)

diff --git a/docs/guide/10-pg_clean.md b/docs/guide/10-pg_clean.md
index 0e11ef0..b528223 100644
--- a/docs/guide/10-pg_clean.md
+++ b/docs/guide/10-pg_clean.md
@@ -1,70 +1,61 @@
-# V2.3.0升级特性pg_clean使用说明
-
-## 1. 安装 pg_clean 工具：
-连接任意主CN：
-
-```
-CREATE EXTENSION pg_clean;
-```
-
-## 2. 构造两阶段事务残留用例：  
-session1—cn1 
- 
-```  
-begin; 
-create table a (id int); 
-prepare transaction 'create_a';
-set xc_maintenance_mode = on;
-execute direct on (dn002) 'rollback prepared ''create_a''';
-set xc_maintenance_mode = off;
-\q
-```
-
-该用例中我们创建了一个gid为 'create\_a' 的两阶段事务，然后将 xc\_maintenance\_mode 打开，使得该事务仅在dn002中执行rollback。
-
-## 3. 查找集群中两阶段事务残留：
-session2—cn2  
-
-```
-select * from pg_clean_check_txn();
-```
-![pg_clean_check_txn](images/v.2.3.0_pg_clean_check_txn.png)
- 
-上图打印出了集群中残留的两阶段事务。其中gid表示该事务的全局标识，global\_transaction\_status表示该事务在全局的状态，transaction\_status\_on\_allnodes表示该事务在所有节点中的状态。
-
-## 4. 查看2PC残留文件名
-
-查询节点cn1中数据目录pg_2pc下所有的2PC残留文件名  
-
-session1—cn1 
-
-```
-	postgres=# select * from pgxc_get_record_list();
-	 pgxc_get_record_list 
-	----------------------
-	 create_a
-	(1 row)
-```
-
-## 5. 清理集群中两阶段事务残留：
-session2—cn2
-
-```
-select * from pg_clean_execute();
-```
-![pg_clean_execute](images/v2.3.0_pg_clean_execute.png)
- 
-上图打印了所有残留的两阶段事务，以及对它们执行的操作。其中operation表示对事务在每个节点执行的操作，operation\_status表示该操作是否执行成功。由于该事务的全局事务状态为ABORT因此，我们去到每个状态为prepare的节点中对该事务执行rollback操作。
-
-## 6. 检查已清理的两阶段事务是否残留文件记录：
-session1—cn1
-
-```
-	postgres=# select * from pgxc_get_record_list();
-	 pgxc_get_record_list 
-	----------------------
- 
-	(1 row)
-```
- 
-由于在上个步骤pg\_clean\_execute执行成功，因此已经回滚的事务'create_a'在所有节点的文件记录均已删除，这里我们在cn1查看所有2pc文件记录，显示为空，结果正确。
+**Upgrade Features of pg_clean V2.3.0 Usage Guide**
+
+1. Install the pg_clean tool:
+   Connect to any primary coordinator (CN):
+
+   ```sql
+   CREATE EXTENSION pg_clean;
+   ```
+
+2. Create a two-phase transaction residue case:
+   Session 1—cn1
+
+   ```sql
+   begin;
+   create table a (id int);
+   prepare transaction 'create_a';
+   set xc_maintenance_mode = on;
+   execute direct on (dn002) 'rollback prepared ''create_a''';
+   set xc_maintenance_mode = off;
+   \q
+   ```
+
+   In this case, a two-phase transaction with a GID of 'create_a' is created. The `xc_maintenance_mode` is then turned on, allowing the transaction to execute a rollback only on dn002.
+
+3. Find residual two-phase transactions in the cluster:
+   Session 2—cn2
+
+   ```sql
+   select * from pg_clean_check_txn();
+   ```
+
+   The output displays residual two-phase transactions in the cluster. GID represents the global identifier of the transaction, global_transaction_status indicates the global status of the transaction, and transaction_status_on_allnodes indicates the status of the transaction on all nodes.
+
+4. View 2PC residue file names:
+   Query all 2PC residue file names in the data directory pg_2pc on node cn1:
+
+   Session 1—cn1
+
+   ```sql
+   select * from pgxc_get_record_list();
+   ```
+
+   The result will show the 'create_a' transaction.
+
+5. Clean up residual two-phase transactions in the cluster:
+   Session 2—cn2
+
+   ```sql
+   select * from pg_clean_execute();
+   ```
+
+   The output displays all residual two-phase transactions and the operations performed on them. The operation column indicates the operation performed on the transaction on each node, and operation_status indicates whether the operation was successful. Since the global transaction status is ABORT, a rollback operation is performed on each node with a prepare status.
+
+6. Check if the cleaned two-phase transactions still have residual file records:
+   Session 1—cn1
+
+   ```sql
+   select * from pgxc_get_record_list();
+   ```
+
+After the successful execution of pg_clean_execute in the previous step, the 'create_a' transaction, which has been rolled back, should have no file records on all nodes. The result should be empty, confirming the correct outcome.