Help Center/ GaussDB(DWS)/ SQL Syntax Reference (9.1.0.x)/ Functions and Operators/ Residual File Management Functions/ Functions for Scanning Residual Files

Updated on 2024-09-30 GMT+08:00

View PDF

Functions for Scanning Residual Files

Only cluster 8.3.0 and later versions support the function of scanning residual files. All functions described in this section can be invoked only by the sysadmin role.

pg_scan_residualfiles()

Description: This function scans all residual file records in the database connected to the current node. It is an instance-level function and is related to the current database, and it can run on any instance.

When it is executed on a CN, it scans the database of the CN and OBS for residual files.
When it is executed on a CN, it scans the database of the DN for residual files.

Return type: record

The following information is displayed:

**Table 1** Return fields
Field	Type	Description
pgscrf	text	Local path of the metadata file that records residual file information.

Example:

    
         SELECT * FROM pg_scan_residualfiles();
                               pgscrf                               
--------------------------------------------------------------------
 pg_residualfiles/pgscrf_meta_1663_16323_20231027143716005244
 pg_residualfiles/pgscrf_meta_2147484281_16323_20231027143716005713
(2 rows)

The restrictions on using this function are as follows:

Only the current database can be scanned for residual files. Currently, only the default tablespace, OBS tablespace, and user-defined tablespace are scanned.
This function is a heavy-load operation and cannot be used concurrently on a single node. You should avoid using this function when the service is overloaded or the resource is scarce.
Only hot partitions in cold-hot-partition-separated tables are supported. Cold partitions are not supported.
Do not use this function during the upgrade observation period.
Residual files in the deleted database cannot be scanned.
Inconsistent cluster metadata information can cause this function incorrectly scan valid data files.

pgxc_scan_residualfiles(query_flag)

Description: Unified CN execution function of pg_scan_residualfiles() This function is a cluster-level function and is related to the current database. It runs on CNs.

Parameter description: query_flag. The parameter type is int, indicating the execution range. 1 indicates the CN, 2 indicates the primary DN, and 4 indicates the standby DN. The query union set can be obtained through the OR operation. For example, 1|2=3 indicates the CN and primary DN, and 1|2|4=7 indicates the CN, primary DN, and standby DN. The default value is 7.

Return type: record

The command output is as follows.

**Table 2** Return fields
Field	Type	Description
nodename	name	Node name.
instance_id	text	Instance name.
pgscrf	text	Local path of the metadata file that records residual file information.

Example:

    
     
       
       
         SELECT * FROM pgxc_scan_residualfiles();
  node_name   | instance_id |                               pgscrf                               
--------------+-------------+--------------------------------------------------------------------
 dn_6001_6002 | dn_6001     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144103839354
 dn_6001_6002 | dn_6001     | pg_residualfiles/pgscrf_meta_2147484281_16323_20231027144103839826
 cn_5001      | cn_5001     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144103946217
 dn_6007_6008 | dn_6008     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144104171311
(4 rows)

SELECT * FROM pgxc_scan_residualfiles(1);
 node_name | instance_id |                            pgscrf                            
-----------+-------------+--------------------------------------------------------------
 cn_5001   | cn_5001     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144421861628
(1 row)

SELECT * FROM pgxc_scan_residualfiles(2);
  node_name   | instance_id |                               pgscrf                               
--------------+-------------+--------------------------------------------------------------------
 dn_6001_6002 | dn_6001     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144424210395
 dn_6001_6002 | dn_6001     | pg_residualfiles/pgscrf_meta_2147484281_16323_20231027144424210855
(2 rows)

SELECT * FROM pgxc_scan_residualfiles(4);
  node_name   | instance_id |                            pgscrf                            
--------------+-------------+--------------------------------------------------------------
 dn_6007_6008 | dn_6008     | pg_residualfiles/pgscrf_meta_1663_16323_20231027144427492060
(1 row)

        

      

    
   

The restrictions on using this function are as follows:

Only the current database can be scanned for residual files. Currently, only the default tablespace, OBS tablespace, and user-defined tablespace are scanned.
This function is a heavy-load operation and cannot be used concurrently on a single node. You should avoid using this function when the service is overloaded or the resource is scarce.
Only hot partitions in cold-hot-partition-separated tables are supported. Cold partitions are not supported.
Do not use this function during the upgrade observation period.
Residual files in the deleted database cannot be scanned.
Inconsistent cluster metadata information can cause this function incorrectly scan valid data files.

pg_get_scan_residualfiles()

Description: Obtains all residual file records scanned on the current node. This function is an instance-level function and is irrelevant to the current database. It can run on any instance.

Return type: record

The command output is as follows.

**Table 3** Return fields
Field	Type	Description
handled	bool	Whether residual files have been processed, moved, or modified.
dbname	text	Database name
residualfile	text	Path of the residual file
size	bigint	File size, in bytes. For residual files in the OBS path, the value of this parameter is 0.
inode	bigint	Inode in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is 0.
atime	timestamptz	Access time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
mtime	timestamptz	Modifie time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
ctime	timestamptz	Change time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
filepath	text	Local path of the metadata file that records residual file information, corresponding to the pgscrf_meta file.
notes	text	Notes

pgxc_get_scan_residualfiles(query_flag)

Description: Unified CN execution function of pg_get_scan_residualfiles() This function is a cluster-level function and is irrelevant to the current database. It runs on CNs.

The query_flag parameter is of the int type and indicates the execution range. 1 indicates the CN, 2 indicates the primary DN, and 4 indicates the standby DN. The query union set can be obtained through the OR operation. For example, 1|2=3 indicates the CN and primary DN, and 1|2|4=7 indicates the CN, primary DN, and standby DN. The default value is 7.

Return type: record

The command output is as follows.

**Table 4** Return fields
Field	Type	Description
nodename	name	Node name
instance_id	text	Instance name
handled	bool	Whether residual files have been processed, moved, or modified.
dbname	text	Database name
residualfile	text	Path of the residual file
size	bigint	File size, in bytes. For residual files in the OBS path, the value of this parameter is 0.
inode	bigint	Inode in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is 0.
atime	timestamptz	Access time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
mtime	timestamptz	Modifie time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
ctime	timestamptz	Change time in the stat information of the residual file. For residual files in the OBS path, the value of this parameter is empty.
filepath	text	Local path of the metadata file that records residual file information, corresponding to the pgscrf_meta file.
notes	text	Notes

pg_archive_scan_residualfiles()

Description: Archives all residual file records scanned on the database connected to the current node. This function is an instance-level function and is related to the current database. It can run on any instance.

Return type: record

The command output is as follows.

**Table 5** Return fields
Field	Type	Description
archive	text	Path of the local folder after archiving. Residual files in the OBS path are archived in the corresponding OBS database directory.
count	bigint	Number of archived residual files.
size	bigint	Total size of archived residual files, in bytes.

Example:

    
         SELECT * FROM pg_archive_scan_residualfiles();
                           archive                            | count | size  
--------------------------------------------------------------+-------+-------
 pg_residualfiles/archive/pgscrf_archive_20231027144613791801 |     4 | 81920
(1 row)

This function can archive only the residual files recorded in the current login database. During archiving, the residual files are verified. The verification results are as follows:
- Verification passed: After the verification is passed, the residual files are archived and marked as handled.
- Verification failed: If the verification fails, the residual files are not archived and marked as handled.
- Verification delayed: If the verification is delayed, it indicates that verification is not possible at the moment, and archiving is skipped. The verification delay is related to the transaction completion status and the standby node's redo progress.
The actual archive directory and the corresponding tablespace are in the same file system. Deleting the tablespace also deletes the archived residual files.
This function cannot be used after DDL delay is enabled.
This function is a heavy-load operation and cannot be used concurrently on a single node. You should avoid using this function when the service is overloaded or the resource is scarce.

pgxc_archive_scan_residualfiles(query_flag)

Description: Unified CN execution function of pg_archive_scan_residualfiles() This function is a cluster-level function and is related to the current database. It runs on CNs.

Return type: record

The command output is as follows.

**Table 6** Return fields
Field	Type	Description
nodename	name	Node name.
instance_id	text	Instance name.
archive	text	Path of the local folder after archiving. Residual files in the OBS path are archived in the corresponding OBS database directory.
count	bigint	Number of archived residual files.
size	bigint	Total size of archived residual files, in bytes.

Example:

    
         SELECT * FROM pgxc_archive_scan_residualfiles();
  node_name   | instance_id |                           archive                            | count | size  
--------------+-------------+--------------------------------------------------------------+-------+-------
 cn_5001      | cn_5001     | pg_residualfiles/archive/pgscrf_archive_20231027145050896440 |     1 | 40960
 dn_6007_6008 | dn_6008     | pg_residualfiles/archive/pgscrf_archive_20231027145051018138 |     1 | 24576
(2 rows)

This function can archive only the residual files recorded in the current login database. During archiving, the residual files are verified. The verification results are as follows:
- Verification passed: After the verification is passed, the residual files are archived and marked as handled.
- Verification failed: If the verification fails, the residual files are not archived and marked as handled.
- Verification delayed: If the verification is delayed, it indicates that verification is not possible at the moment, and archiving is skipped. The verification delay is related to the transaction completion status and the standby node's redo progress.
The actual archive directory and the corresponding tablespace are in the same file system. Deleting the tablespace also deletes the archived residual files.
This function cannot be used after DDL delay is enabled.
This function is a heavy-load operation and cannot be used concurrently on a single node. You should avoid using this function when the service is overloaded or the resource is scarce.

pg_rm_scan_residualfiles_archive()

Description: Deletes all archived residual file records on the current node. This function is an instance-level function and is irrelevant to the current database. It can run on any instance.

Return type: record

The command output is as follows.

**Table 7** Return fields
Field	Type	Description
count	bigint	Number of residual files that have been deleted from the archive. For residual files in the local path, the number of deleted files is counted. For residual files in the OBS path, the number of deleted table directories is counted.
size	bigint	Total size of residual files that have been deleted from the archive, in bytes. For residual files in the OBS path, the value of this parameter is 0.

Example:

    
         SELECT * FROM pg_rm_scan_residualfiles_archive();
 count | size  
-------+-------
     4 | 81920
(1 row)

pgxc_rm_scan_residualfiles_archive(query_flag)

Description: Unified CN execution function of pg_rm_scan_residualfiles_archive() This function is a cluster-level function and is irrelevant to the current database. It runs on CNs.

Return type: record

The command output is as follows.

**Table 8** Return fields
Field	Type	Description
nodename	name	Node name.
instance_id	text	Instance name.
count	bigint	Number of residual files that have been deleted from the archive. For residual files in the local path, the number of deleted files is counted. For residual files in the OBS path, the number of deleted table directories is counted.
size	bigint	Total size of residual files that have been deleted from the archive, in bytes. For residual files in the OBS path, the value of this parameter is 0.

Example:

    
         SELECT * FROM pgxc_rm_scan_residualfiles_archive();
  node_name   | instance_id | count | size  
--------------+-------------+-------+-------
 dn_6001_6002 | dn_6001     |     4 | 81920
 cn_5001      | cn_5001     |     1 | 40960
 dn_6007_6008 | dn_6008     |     1 | 24576
 coordinator1 | coordinator1|     1 |    0 
(4 rows)