openGauss增量备份恢复

openGauss 增量备份恢复

openGauss 数据库自 2020 年 6 月 30 日发布以来，很多小伙伴都提到“openGauss 数据库是否有增量备份工具？“这么一个问题。

在 openGauss 1.0.0 版本的时候，关于这个问题的回答往往是：“Sorry…”，openGauss 数据库可以使用 gs_basebackup 工具对数据库进行物理全备份，然后通过对 WAL 日志的备份间接实现数据库的增量备份。

当 openGauss 1.1.0 版本于 2020 年 12 月 31 日发布后，有小伙伴说 openGauss 可以使用 gs_probackup 实现数据库的增量备份。(实际上该工具应该是在 1.0.1 版本已经发布)

本人在 openGauss 的《管理员指南》文档中并没有看到该工具，但在《openGauss 工具参考》文档的“系统内部使用的工具”章节看到了该工具的介绍。于是对该工具的使用做了简单测试，并将相关操作总结如下：

  • 常见功能：

    • 可用于备份单机数据库或者主节点数据库，为物理备份；
    • 可备份外部目录的内容，如脚本文件、配置文件、日志文件、dump 文件等；
    • 支持增量备份、定期备份和远程备份； 可设置备份的留存策略。

  • Tips：

    若要使用 PTRACK 增量备份，需在 postgresql.conf 中手动添加参数“enable_cbm_tracking = on”。

    openGauss 可以动态跟踪页面更改。在每次更新数据页时，openGauss 都会在 PTRACK 位图中进行相应的标记。由于一个页面只需要 PTRACK fork 中的一个位，这样的位图非常小。

    跟踪意味着数据库服务器操作上的一些小开销，但会显著加快增量备份。PTRACK 的另一个优点是它为增量备份提供了页面级的粒度。

  • 限制说明：

    • 备份必须由运行数据库服务器的用户执行；
    • 备份和恢复的数据库服务器的主版本号必须相同；
    • 如果通过 ssh 在远程模式下备份数据库，需要在本地和远程主机安装相同主版本的数据库且建立 SSH 互信；
    • 远程模式下只能执行 add-instance、backup、restore 子命令；
    • 使用 restore 子命令前，应先停止 gaussdb 进程。

  • 常见的操作类型：

备份流程示例

1. 打开参数 enable_cbm_tracking,跟踪数据页的变化

mydb=# show enable_cbm_tracking;enable_cbm_tracking
---------------------offmydb=# alter system set enable_cbm_tracking=on;
ALTER SYSTEM SET
mydb=# show enable_cbm_tracking;enable_cbm_tracking
---------------------on

2. 初始化备份目录(/home/omm/gs_bak2021)

[omm@prod ~]$ gs_probackup --version
gs_probackup (openGauss 1.1.0 build 392c0438) compiled at 2020-12-31 20:07:42 commit 0 last mr[omm@prod ~]$ gs_probackup init -B /home/omm/gs_bak2021/
INFO: Backup catalog '/home/omm/gs_bak2021' successfully inited
[omm@prod ~]$ tree -L 3 /home/omm/gs_bak2021/
/home/omm/gs_bak2021/
├── backups
└── wal2 directories, 0 files## 对目录的初始化操作实际是在备份目录下创建backups/和wal/子目录，分别用于存放备份文件和WAL文件。

3. 添加备份实例

  • add-instance 语法：
gs_probackup add-instance -B backup-path                       ## 备份路径-D pgdata-path                       ## 数据目录路径--instance=instance_name             ## 实例名(自定义)[-E external-directories-paths]      ## 需要备份的其他目录[--remote-proto=protocol]            ## 远程操作的协议(仅支持SSH)[--remote-host=destination]          ## 远程主机IP或主机名[--remote-path=path]                 ## 远程主机的gs_probackup安装目录[--remote-user=username]             ## 远程主机SSH连接的用户[--remote-port=port]                 ## 远程主机SSH连接的端口(默认22)[--ssh-options=ssh_options]          ## SSH命令行参数[--help]                             ## 显示帮助信息
  • 执行示例：
[omm@prod ~]$ gs_probackup add-instance -B /home/omm/gs_bak2021 -D /gauss/data/db1 --instance gs_bak2021_inst
INFO: Instance 'gs_bak2021_inst' successfully inited[omm@prod ~]$ gs_probackup show -B /home/omm/gs_bak2021/
BACKUP INSTANCE 'gs_bak2021_inst'
=================================================================================================================Instance  Version  ID  Recovery Time  Mode  WAL Mode  TLI  Time  Data  WAL  Zratio  Start LSN  Stop LSN  Status
=================================================================================================================

4. 执行一次全量备份

  • backup 语法：
gs_probackup backup -B backup-path --instance=instance_name [-D pgdata-path]-b backup-mode                                       ## 备份模式：FULL(全量备份)/PTRACK(增量备份)[-C]                                                 ## -C指定检查点模式为spread(周期调度)，默认为fast快速完成[-S slot-name] [--temp-slot]                         ## -S指定物理复制slot名称(默认pg_probackup_slot) ,--temp-slot为WAL流处理创建临时物理复制slot，确保备份过程中所需的WAL段仍然可用[--backup-pg-log]                                    ## 备份日志目录(默认不备份)[-j threads_num]                                     ## 并行线程数[--progress]                                         ## 显示进度[--no-validate]                                      ## 完成备份后跳过自动验证[--skip-block-validation]                            ## 关闭块级校验，加快备份速度[-E external-directories-paths]                      ## 需要备份的其他目录[--no-sync]                                          ## 不将备份文件同步写入磁盘，提升写入性能[--note=text]                                        ## 备份的注释[--archive-timeout=timeout]                          ## 以秒为单位设置流式处理的超时时间(默认300)[--log-level-console=log-level-console]              ## 设置要发送到控制台的日志级别，默认为info，设置off可以关闭[--log-level-file=log-level-file]                    ## 设置要发送到日志文件的日志级别，默认off[--log-filename=log-filename]                        ## 指定要创建的日志文件的文件名(strftime模式示例：pg_probackup-%u.log),默认pg_probackup.log[--error-log-filename=error-log-filename]            ## error日志的日志文件名[--log-directory=log-directory]                      ## gs_probackup的日志目录[--log-rotation-size=log-rotation-size]              ## gs_probackup的日志大小(默认单位KB,支持KB、MB、GB、TB)，超过阈值后进行循环，默认0(禁用)[--log-rotation-age=log-rotation-age]                ## 单个日志文件的最大生命周期(默认单位min,支持ms, s, min, h, d)，默认0表示禁用基于时间的循环[--delete-expired]                                   ## 删除不符合pg_probackup.conf配置文件中定义的留存策略的备份[--delete-wal]                                       ## 删除不需要的WAL文件[--merge-expired]                                    ## 将满足留存策略要求的最旧的增量备份与其已过期的父备份合并[--retention-redundancy=retention-redundancy]        ## 留存的完整备份数(默认0,禁用此设置)[--retention-window=retention-window]                ## 留存的天数(默认0,禁用此设置)[--wal-depth=wal-depth]                              ## 每个时间轴上必须留存的执行PITR能力的最新有效备份数(默认0,禁用此设置)[--dry-run]                                          ## 显示所有可用备份的当前状态，不删除或合并过期备份[--ttl=interval]                                     ## 从恢复时间开始计算，备份要固定的时间量(支持单位：ms, s, min, h, d(默认为s))[--expire-time=time]                                 ## 备份固定失效的时间戳(例如：--expire-time='2020-01-01 00:00:00+03')[--compress-algorithm=compress-algorithm][--compress-level=compress-level][--compress]   ## 压缩参数：压缩算法(zlib/pglz/none)/压缩级别(0~9,默认1)/压缩(相当于zlib算法+1压缩级别)[-d dbname] [-h host] [-p port] [-U username] [-w] [-W password]                         ## 连接信息：数据库名/主机名/端口/用户/密码[--remote-proto=protocol] [--remote-host=destination][--remote-path=path] [--remote-user=username][--remote-port=port] [--ssh-options=ssh_options][--help]
  • 执行示例：
[omm@prod ~]$ gs_probackup backup -B /home/omm/gs_bak2021 --instance gs_bak2021_inst -b full -D /gauss/data/db1 -d mydb -p 26000 --progress \
> --log-directory=/home/omm/gs_bak2021/log  --log-rotation-size=10GB --log-rotation-age=30d  --log-level-file=info  --log-filename=full_20210111.log \
> --retention-redundancy=2 \
> --compress  \
> --note='This is full backup set.'INFO: Backup start, gs_probackup version: 2.4.2, instance: gs_bak2021_inst, backup ID: QMR0ZE, backup mode: FULL, wal mode: STREAM, remote: false, compress-algorithm: zlib, compress-level: 1
LOG: Backup destination is initialized
WARNING: This openGauss instance was initialized without data block checksums. gs_probackup have no way to detect data block corruption without them. Reinitialize PGDATA with option '--data-checksums'.
INFO: Adding note to backup QMR0ZE: 'This is full backup set.'
LOG: Database backup start
INFO: Cannot parse path "base"
INFO: PGDATA size: 576MB
LOG: started streaming WAL at 0/F000000 (timeline 1)check identify system successsend START_REPLICATION 0/F000000 successkeepalive message is receivedkeepalive message is received
INFO: Start transferring data files
INFO: Progress: (1/1652). Process file "postgresql.conf.bak"
... ...
INFO: Progress: (1652/1652). Process file "base/16388/14815"
INFO: Data files are transferred, time elapsed: 3s
INFO: wait for pg_stop_backup()keepalive message is received
INFO: pg_stop backup() successfully executed
LOG: stop_lsn: 0/F0001E0
LOG: Looking for LSN 0/F0001E0 in segment: 00000001000000000000000F
LOG: Found WAL segment: /home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR0ZE/database/pg_xlog/00000001000000000000000F
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR0ZE/database/pg_xlog/00000001000000000000000F"
LOG: Found LSN: 0/F0001E0
(null): not renaming 000000010000000000000010, segment is not complete.
LOG: finished streaming WAL at 0/10000140 (timeline 1)
LOG: Getting the Recovery Time from WAL
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR0ZE/database/pg_xlog/00000001000000000000000F"
INFO: Syncing backup files to disk
INFO: Backup files are synced, time elapsed: 1s
INFO: Validating backup QMR0ZE
INFO: Progress: (1/1655). Validate file "postgresql.conf.bak"
... ...
INFO: Progress: (1655/1655). Validate file "database_map"
INFO: Backup QMR0ZE data files are valid
INFO: Backup QMR0ZE resident size: 567MB
INFO: Backup QMR0ZE completed[omm@prod ~]$ gs_probackup show -B /home/omm/gs_bak2021/BACKUP INSTANCE 'gs_bak2021_inst'
========================================================================================================================================Instance         Version  ID      Recovery Time           Mode  WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN  Stop LSN   Status
========================================================================================================================================gs_bak2021_inst  9.2      QMR0ZE  2021-01-11 10:45:17+08  FULL  STREAM    1/0    9s  551MB  16MB    1.05  0/F000028  0/F0001E0  OK

5. 执行增量备份

  • 创建测试数据 1
mydb=# create table incr_bak1(name varchar(50));
CREATE TABLE
mydb=# insert into incr_bak1 values('This is the first change.');
INSERT 0 1
mydb=# select * from incr_bak1;name
---------------------------This is the first change.
  • 执行第一次增量备份
[omm@prod ~]$ gs_probackup backup -B /home/omm/gs_bak2021 --instance gs_bak2021_inst -b PTRACK -D /gauss/data/db1 -d mydb -p 26000 --progress \
> --log-directory=/home/omm/gs_bak2021/log  --log-rotation-size=10GB --log-rotation-age=30d  --log-level-file=info  --log-filename=incr1_20210111.log \
> --delete-expired --delete-wal \
> --retention-redundancy=2 \
> --compress  \
> --note='This is the first incremental backup set.'
INFO: Backup start, gs_probackup version: 2.4.2, instance: gs_bak2021_inst, backup ID: QMR19V, backup mode: PTRACK, wal mode: STREAM, remote: false, compress-algorithm: zlib, compress-level: 1
LOG: Backup destination is initialized
WARNING: This openGauss instance was initialized without data block checksums. gs_probackup have no way to detect data block corruption without them. Reinitialize PGDATA with option '--data-checksums'.
INFO: Adding note to backup QMR19V: 'This is the first incremental backup set.'
LOG: Database backup start
LOG: Latest valid FULL backup: QMR0ZE
INFO: Parent backup: QMR0ZE
INFO: Cannot parse path "base"
INFO: PGDATA size: 576MB
LOG: Current tli: 1
LOG: Parent start_lsn: 0/F000028
LOG: start_lsn: 0/11000028
INFO: Extracting pagemap of changed blocks
LOG: started streaming WAL at 0/11000000 (timeline 1)check identify system successsend START_REPLICATION 0/11000000 successkeepalive message is receivedkeepalive message is received
INFO: Pagemap successfully extracted, time elapsed: 0 sec
INFO: Start transferring data files
INFO: Progress: (1/1653). Process file "postgresql.conf.bak"
... ...
INFO: Progress: (1653/1653). Process file "base/16388/14737_vm"
INFO: Data files are transferred, time elapsed: 0keepalive message is received
INFO: wait for pg_stop_backup()
INFO: pg_stop backup() successfully executed
LOG: stop_lsn: 0/110002F8
LOG: Looking for LSN 0/110002F8 in segment: 000000010000000000000011
LOG: Found WAL segment: /home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR19V/database/pg_xlog/000000010000000000000011
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR19V/database/pg_xlog/000000010000000000000011"
LOG: Found LSN: 0/110002F8
(null): not renaming 000000010000000000000012, segment is not complete.
LOG: finished streaming WAL at 0/12000140 (timeline 1)
LOG: Getting the Recovery Time from WAL
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR19V/database/pg_xlog/000000010000000000000011"
INFO: Syncing backup files to disk
INFO: Backup files are synced, time elapsed: 1s
INFO: Validating backup QMR19V
INFO: Progress: (1/1656). Validate file "postgresql.conf.bak"
... ...
INFO: Progress: (1656/1656). Validate file "database_map"
INFO: Backup QMR19V data files are valid
INFO: Backup QMR19V resident size: 289MB
INFO: Backup QMR19V completed
LOG: REDUNDANCY=2
INFO: Evaluate backups by retention
INFO: Backup QMR19V, mode: PTRACK, status: OK. Redundancy: 1/2, Time Window: 0d/0d. Active
INFO: Backup QMR0ZE, mode: FULL, status: OK. Redundancy: 1/2, Time Window: 0d/0d. Active
INFO: There are no backups to merge by retention policy
INFO: There are no backups to delete by retention policy
INFO: There is no WAL to purge by retention policy[omm@prod gs_bak2021]$ gs_probackup show -B /home/omm/gs_bak2021/BACKUP INSTANCE 'gs_bak2021_inst'
============================================================================================================================================Instance         Version  ID      Recovery Time           Mode    WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN   Stop LSN    Status
============================================================================================================================================gs_bak2021_inst  9.2      QMR19V  2021-01-11 10:51:31+08  PTRACK  STREAM    1/1    5s  273MB  16MB    0.94  0/11000028  0/110002F8  OKgs_bak2021_inst  9.2      QMR0ZE  2021-01-11 10:45:17+08  FULL    STREAM    1/0    9s  551MB  16MB    1.05  0/F000028   0/F0001E0   OK
  • 创建测试数据 2
mydb=# create table incr_bak2(name varchar(50));
CREATE TABLE
mydb=# insert into incr_bak2 values('This is the second change.');
INSERT 0 1
mydb=# select * from incr_bak2;name
----------------------------This is the second change.
  • 执行第二次增量备份
[omm@prod ~]$ gs_probackup backup -B /home/omm/gs_bak2021 --instance gs_bak2021_inst -b PTRACK -D /gauss/data/db1 -d mydb -p 26000 --progress \
> --log-directory=/home/omm/gs_bak2021/log  --log-rotation-size=10GB --log-rotation-age=30d  --log-level-file=info  --log-filename=incr2_20210111.log \
> --delete-expired --delete-wal \
> --retention-redundancy=2 \
> --compress  \
> --note='This is the second incremental backup set.'
INFO: Backup start, gs_probackup version: 2.4.2, instance: gs_bak2021_inst, backup ID: QMR1J4, backup mode: PTRACK, wal mode: STREAM, remote: false, compress-algorithm: zlib, compress-level: 1
LOG: Backup destination is initialized
WARNING: This openGauss instance was initialized without data block checksums. gs_probackup have no way to detect data block corruption without them. Reinitialize PGDATA with option '--data-checksums'.
INFO: Adding note to backup QMR1J4: 'This is the second incremental backup set.'
LOG: Database backup start
LOG: Latest valid FULL backup: QMR0ZE
INFO: Parent backup: QMR19V
INFO: Cannot parse path "base"
INFO: PGDATA size: 576MB
LOG: Current tli: 1
LOG: Parent start_lsn: 0/11000028
LOG: start_lsn: 0/15000028
INFO: Extracting pagemap of changed blocks
LOG: started streaming WAL at 0/15000000 (timeline 1)check identify system successsend START_REPLICATION 0/15000000 successkeepalive message is receivedkeepalive message is received
INFO: Pagemap successfully extracted, time elapsed: 0 sec
INFO: Start transferring data files
INFO: Progress: (1/1654). Process file "postgresql.conf.bak"
... ...(开始传输数据)
INFO: Progress: (1654/1654). Process file "base/16388/15093"
INFO: Data files are transferred, time elapsed: 1skeepalive message is received
INFO: wait for pg_stop_backup()
INFO: pg_stop backup() successfully executed
LOG: stop_lsn: 0/150002F8
LOG: Looking for LSN 0/150002F8 in segment: 000000010000000000000015
LOG: Found WAL segment: /home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR1J4/database/pg_xlog/000000010000000000000015
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR1J4/database/pg_xlog/000000010000000000000015"
LOG: Found LSN: 0/150002F8
(null): not renaming 000000010000000000000016, segment is not complete.
LOG: finished streaming WAL at 0/16000140 (timeline 1)
LOG: Getting the Recovery Time from WAL
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMR1J4/database/pg_xlog/000000010000000000000015"
INFO: Syncing backup files to disk
INFO: Backup files are synced, time elapsed: 1s
INFO: Validating backup QMR1J4
... ...(验证备份有效性)
INFO: Backup QMR1J4 data files are valid
INFO: Backup QMR1J4 resident size: 289MB
INFO: Backup QMR1J4 completed
LOG: REDUNDANCY=2
INFO: Evaluate backups by retention
INFO: Backup QMR1J4, mode: PTRACK, status: OK. Redundancy: 1/2, Time Window: 0d/0d. Active
INFO: Backup QMR19V, mode: PTRACK, status: OK. Redundancy: 1/2, Time Window: 0d/0d. Active
INFO: Backup QMR0ZE, mode: FULL, status: OK. Redundancy: 1/2, Time Window: 0d/0d. Active
INFO: There are no backups to merge by retention policy
INFO: There are no backups to delete by retention policy
INFO: There is no WAL to purge by retention policy[omm@prod ~]$ gs_probackup show -B /home/omm/gs_bak2021/BACKUP INSTANCE 'gs_bak2021_inst'
============================================================================================================================================Instance         Version  ID      Recovery Time           Mode    WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN   Stop LSN    Status
============================================================================================================================================gs_bak2021_inst  9.2      QMR1J4  2021-01-11 10:57:05+08  PTRACK  STREAM    1/1    5s  273MB  16MB    0.94  0/15000028  0/150002F8  OKgs_bak2021_inst  9.2      QMR19V  2021-01-11 10:51:31+08  PTRACK  STREAM    1/1    5s  273MB  16MB    0.94  0/11000028  0/110002F8  OKgs_bak2021_inst  9.2      QMR0ZE  2021-01-11 10:45:17+08  FULL    STREAM    1/0    9s  551MB  16MB    1.05  0/F000028   0/F0001E0   OK

6. 模拟“删库”，但不跑路

-- 删除数据库mydb
postgres=# drop database mydb;
DROP DATABASE-- 删除表t1(数据库postgres)
postgres=# \dList of relationsSchema | Name | Type  | Owner |             Storage
--------+------+-------+-------+----------------------------------public | t1   | table | omm   | {orientation=row,compression=no}public | test | table | omm   | {orientation=row,compression=no}postgres=# drop table t1;
DROP TABLE
postgres=# \dList of relationsSchema | Name | Type  | Owner |             Storage
--------+------+-------+-------+----------------------------------public | test | table | omm   | {orientation=row,compression=no}

7. 增量恢复测试

  • restore 语法：
gs_probackup restore -B backup-path --instance=instance_name[-D pgdata-path] [-i backup-id] [-j threads_num] [--progress][--force]                                 ## 允许从损坏的或无效的备份中恢复数据(慎用)[--no-sync]                               ## 不同步写入磁盘,以提升恢复效率[--no-validate] [--skip-block-validation] ## 跳过备份验证/跳过块级验证(仅做文件级别验证)[--external-mapping=OLDDIR=NEWDIR]        ## 恢复时重新指定目录OLDIR为NEWDIR[-T OLDDIR=NEWDIR]                        ## 与--external-mapping配合使用,重新指定表空间路径(将表空间路径OLDDIR改为NEWDIR)[--skip-external-dirs]                    ## 不恢复备份文件中--external-dirs选项所指定的外部目录[-I incremental_mode]                     ## 增量模式none|checksum|lsn,默认none[--recovery-target-time=time              ## 指定恢复的目标time(只能指定备份信息内的recovery-time) , 备份信息使用gs_probackup show查看|--recovery-target-xid=xid                ## 指定恢复的目标xid(只能指定备份信息内的recovery-xid)|--recovery-target-lsn=lsn                ## 指定恢复的目标lsn(只能指定备份信息内的stop-lsn)|--recovery-target-name=target-name]      ## 指定恢复的目标保存点(只能指定备份信息内的recovery-name)[--recovery-target-inclusive=boolean]     ## true(恢复目标将包括指定的内容) | false(恢复目标将不包括指定的内容),与以上recovery-target配合使用[--remote-proto=protocol] [--remote-host=destination][--remote-path=path] [--remote-user=username][--remote-port=port] [--ssh-options=ssh_options] ## 远程连接信息[--log-level-console=log-level-console][--log-level-file=log-level-file][--log-filename=log-filename][--error-log-filename=error-log-filename][--log-directory=log-directory][--log-rotation-size=log-rotation-size][--log-rotation-age=log-rotation-age][--help]
## 其他参数介绍参考前面的 gs_probackup backup语法介绍.

删除整个数据库目录(database cluster)

[omm@prod ~]$ gs_ctl stop -D /gauss/data/db1/
[2021-01-11 11:47:39.259][2211][][gs_ctl]: gs_ctl stopped ,datadir is /gauss/data/db1
waiting for server to shut down............. done
server stopped
[omm@prod ~]$ rm -fr /gauss/data/db1/

执行全量恢复并验证数据

## 全量恢复
[omm@prod ~]$ gs_probackup restore -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -D /gauss/data/db1 -i QMR0ZE --progress -j 4    ##  -i指定备份文件ID，QMR0ZE即为全备的ID
LOG: Restore begin.
LOG: check tablespace directories of backup QMR0ZE
LOG: check external directories of backup QMR0ZE
... ...
大致流程：
1. 校验备份集QMR0ZE的有效性
2. 校验备份集QMR0ZE中WAL segments的有效性
3. 开始恢复数据(Restoring the database from backup at 2021-01-11 10:45:14+08)
INFO: Start restoring backup files. PGDATA size: 592MB
LOG: Start thread 1
LOG: Start thread 2
LOG: Start thread 3
LOG: Start thread 4
... ...
INFO: Syncing restored files to disk
INFO: Restored backup files are synced, time elapsed: 5s
INFO: Restore of backup QMR0ZE completed.## 验证数据正常(全备时刻，不包含表incr_bak1和incr_bak2)
[omm@prod ~]$ gs_ctl start -D /gauss/data/db1
[omm@prod ~]$ gsql -d mydb -p 26000 -r
gsql ((openGauss 1.1.0 build 392c0438) compiled at 2020-12-31 20:07:42 commit 0 last mr  )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.mydb=# \dList of relationsSchema |   Name   |       Type        | Owner |             Storage
--------+----------+-------------------+-------+----------------------------------public | dept     | table             | omm   | {orientation=row,compression=no}public | emp      | table             | omm   | {orientation=row,compression=no}public | mv_emp   | materialized view | omm   |public | salgrade | table             | omm   | {orientation=row,compression=no}public | seq1     | sequence          | omm   |public | v_emp    | view              | omm   |

执行全量+第一次增量恢复测试并验证数据( restore 操作要求数据目录必须为空 )

[omm@prod ~]$ gs_probackup restore -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -D /gauss/data/db1 -i QMR19V --progress  ##  -i指定备份文件ID，QMR19V即第一次增备的ID
LOG: Restore begin.
LOG: check tablespace directories of backup QMR19V
LOG: check external directories of backup QMR19V
... ...
大致流程：
1. 校验备份集(QMR0ZE/QMR19V)的有效性;
2. 验证备份集QMR19V中的WAL segments的有效性(本次测试验证的WAL日志是000000010000000000000011);
3. 开始恢复数据库(Restoring the database from backup at 2021-01-11 10:51:31+08)
... ...
INFO: Syncing restored files to disk
INFO: Restored backup files are synced, time elapsed: 5s
INFO: Restore of backup QMR19V completed.## 验证数据正常(全备+1次增备包含表incr_bak1,但不包含表incr_bak2)
[omm@prod ~]$ gs_ctl start -D /gauss/data/db1/
[omm@prod ~]$ gsql -d mydb -p 26000 -r
gsql ((openGauss 1.1.0 build 392c0438) compiled at 2020-12-31 20:07:42 commit 0 last mr  )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.mydb=# \dList of relationsSchema |   Name    |       Type        | Owner |             Storage
--------+-----------+-------------------+-------+----------------------------------public | dept      | table             | omm   | {orientation=row,compression=no}public | emp       | table             | omm   | {orientation=row,compression=no}public | incr_bak1 | table             | omm   | {orientation=row,compression=no}public | mv_emp    | materialized view | omm   |public | salgrade  | table             | omm   | {orientation=row,compression=no}public | seq1      | sequence          | omm   |public | v_emp     | view              | omm   |mydb=# select * from incr_bak1;name
---------------------------This is the first change.

执行全量+两次增量恢复测试并验证数据( restore 操作要求数据目录必须为空 )

[omm@prod ~]$ gs_probackup restore -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -D /gauss/data/db1 -i QMR1J4 --progress   ##  -i指定备份文件ID，QMR1J4即第二次增备的ID
LOG: Restore begin.
LOG: check tablespace directories of backup QMR1J4
LOG: check external directories of backup QMR1J4
INFO: Validating parents for backup QMR1J4
WARNING: Process 2449 which used backup QMR0ZE no longer exists
... ...
大致流程：
1. 校验备份集(QMR0ZE/QMR19V/QMR1J4)的有效性;
2. 还原备份集(Restoring the database from backup at 2021-01-11 10:57:04+08)
... ...
INFO: Syncing restored files to disk
INFO: Restored backup files are synced, time elapsed: 4s
INFO: Restore of backup QMR1J4 completed.## 验证数据正常(全备+2次增备包含表incr_bak1和incr_bak2)
[omm@prod ~]$ gs_ctl start -D /gauss/data/db1/
[omm@prod ~]$ gsql -d mydb -p 26000 -r
gsql ((openGauss 1.1.0 build 392c0438) compiled at 2020-12-31 20:07:42 commit 0 last mr  )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.mydb=# \dList of relationsSchema |   Name    |       Type        | Owner |             Storage
--------+-----------+-------------------+-------+----------------------------------public | dept      | table             | omm   | {orientation=row,compression=no}public | emp       | table             | omm   | {orientation=row,compression=no}public | incr_bak1 | table             | omm   | {orientation=row,compression=no}public | incr_bak2 | table             | omm   | {orientation=row,compression=no}public | mv_emp    | materialized view | omm   |public | salgrade  | table             | omm   | {orientation=row,compression=no}public | seq1      | sequence          | omm   |public | v_emp     | view              | omm   |mydb=# select * from incr_bak1,incr_bak2;name            |            name
---------------------------+----------------------------This is the first change. | This is the second change.

其他示例

  • 备份集详细信息查询
[omm@prod log]$ gs_probackup show -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -i QMR0ZE
#Configuration
backup-mode = FULL
stream = true
compress-alg = zlib
compress-level = 1
from-replica = false#Compatibility
block-size = 8192
xlog-block-size = 8192
checksum-version = 0
program-version = 2.4.2
server-version = 9.2#Result backup info
timelineid = 1
start-lsn = 0/F000028
stop-lsn = 0/F0001E0
start-time = '2021-01-11 10:45:14+08'
end-time = '2021-01-11 10:45:23+08'
recovery-xid = 16547
recovery-time = '2021-01-11 10:45:17+08'
recovery-name = 'backup QMR0ZE'
data-bytes = 577690814
wal-bytes = 16777216
uncompressed-bytes = 604023922
pgdata-bytes = 604023922
status = OK
note = 'This is full backup set.'
content-crc = 825608417
  • 删除备份实例
## 语法详见：gs_probackup del-instance --help
gs_probackup del-instance -B /home/omm/gs_bak2021/ -D /gauss/data/db1/ --instance=gs_bak2021_inst
  • 删除备份集
## 语法详见：gs_probackup delete --help
gs_probackup delete -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -i QMR0OF
  • 合并备份集 (语法介绍：gs_probackup merge --help)
[omm@prod ~]$ gs_probackup show -B /home/omm/gs_bak2021/BACKUP INSTANCE 'gs_bak2021_inst'
============================================================================================================================================Instance         Version  ID      Recovery Time           Mode    WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN   Stop LSN    Status
============================================================================================================================================gs_bak2021_inst  9.2      QMR1J4  2021-01-11 10:57:05+08  PTRACK  STREAM    1/1    5s  273MB  16MB    0.94  0/15000028  0/150002F8  OKgs_bak2021_inst  9.2      QMR19V  2021-01-11 10:51:31+08  PTRACK  STREAM    1/1    5s  273MB  16MB    0.94  0/11000028  0/110002F8  OKgs_bak2021_inst  9.2      QMR0ZE  2021-01-11 10:45:17+08  FULL    STREAM    1/0    9s  551MB  16MB    1.05  0/F000028   0/F0001E0   OK[omm@prod ~]$ gs_probackup merge -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -i QMR1J4 --progress
INFO: Merge started
WARNING: Process 2591 which used backup QMR0ZE no longer exists
WARNING: Process 2591 which used backup QMR19V no longer exists
WARNING: Process 2527 which used backup QMR1J4 no longer exists
INFO: Merging backup QMR1J4 with parent chain
... ...
大致流程：
1. 校验备份集QMR0ZE/QMR19V/QMR1J4
2. 合并备份集QMR1J4和QMR19V至备份集QMR0ZE(合并至全备)
3. 删除备份集QMR19V和QMR1J4(删除增备)
4. 重命名QMR0ZE为QMR1J4 (将原先全备的备份ID重命名为最近的刚被删除的那个增备ID：QMR1J4)
5. 校验新的备份集QMR1J4数据文件是否有效
... ...
INFO: Merge of backup QMR1J4 completed## 合并结束 (新备份集QMR1J4与原备份集QMR1J4的备份信息相同)
[omm@prod ~]$ gs_probackup show -B /home/omm/gs_bak2021/BACKUP INSTANCE 'gs_bak2021_inst'
==========================================================================================================================================Instance         Version  ID      Recovery Time           Mode  WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN   Stop LSN    Status
==========================================================================================================================================gs_bak2021_inst  9.2      QMR1J4  2021-01-11 10:57:05+08  FULL  STREAM    1/0   10s  829MB  16MB    1.10  0/15000028  0/150002F8  OK
  • 验证备份集是否有效 (语法介绍：gs_probackup validate --help)
[omm@prod database]$ gs_probackup validate -B /home/omm/gs_bak2021/ --instance=gs_bak2021_inst -i QMRGCU
LOG: Validate begin.
INFO: Validating parents for backup QMRGCU
WARNING: Process 2864 which used backup QMR1J4 no longer exists
INFO: Validating backup QMR1J4
INFO: Backup QMR1J4 data files are valid
WARNING: Process 2864 which used backup QMRGCU no longer exists
INFO: Validating backup QMRGCU
INFO: Backup QMRGCU data files are valid
LOG: Thread [1]: Opening WAL segment "/home/omm/gs_bak2021/backups/gs_bak2021_inst/QMRGCU/database/pg_xlog/000000010000000000000019"
INFO: Backup QMRGCU WAL segments are valid
INFO: Backup QMRGCU is valid.
INFO: Validate of backup QMRGCU completed.
  • gs_probackup 配置
## 配置文件主要包含2类：
# 1. 备份实例目录下的pg_probackup.conf文件(指定备份时使用的相关参数)
# 2. 备份集目录下的backup.control文件(描述备份集的属性信息)# 语法：
gs_probackup set-backup -B backup-path --instance=instance_name -i backup-id[--note=text]         ## 备份集的注释[--ttl=interval]      ## 备份集有效期(单位： 'ms', 's', 'min', 'h', 'd' (默认: s))    -- 与--expire-time互斥[--expire-time=time]  ## 备份集的失效日期(如：'2021-10-01 00:00:00+08')               -- 与--ttl互斥gs_probackup set-config -B backup-path --instance=instance_name[-D pgdata-path] [-E external-directories-paths][--archive-timeout=timeout][--retention-redundancy=retention-redundancy][--retention-window=retention-window][--wal-depth=wal-depth][--compress-algorithm=compress-algorithm][--compress-level=compress-level][-d dbname] [-h host] [-p port] [-U username][--log-level-console=log-level-console][--log-level-file=log-level-file][--log-filename=log-filename][--error-log-filename=error-log-filename][--log-directory=log-directory][--log-rotation-size=log-rotation-size][--log-rotation-age=log-rotation-age][--remote-proto=protocol] [--remote-host=destination][--remote-path=path] [--remote-user=username][--remote-port=port] [--ssh-options=ssh_options]
## gs_probackup set-config 的参数均是前面备份时使用的参数，这里就不做详细解释## 示例：
[omm@prod ~]$ gs_probackup set-backup -B /home/omm/gs_bak2021 --instance=gs_bak2021_inst -i QMRIA3 --note=incr_bak2 --ttl=10d
INFO: Backup QMRIA3 is pinned until '2021-01-21 16:58:52+08'
INFO: Adding note to backup QMRIA3: 'incr_bak2'[omm@prod ~]$ gs_probackup set-config -B /home/omm/gs_bak2021 --instance=gs_bak2021_inst -D /gauss/data/db1 \
> --retention-redundancy=2 --compress-algorithm=zlib --compress-level=6 \
> --log-filename='pg_probackup-%Y-%m-%d_%H%M%S.log' \
> --log-directory=/home/omm/gs_bak2021 --log-rotation-size=1GB## 查看配置文件
[omm@prod ~]$ cat /home/omm/gs_bak2021/backups/gs_bak2021_inst/pg_probackup.conf
# Backup instance information
pgdata = /gauss/data/db1
system-identifier = 18020659335453983985
# Logging parameters
log-filename = pg_probackup-%Y-%m-%d_%H%M%S.log
log-directory = /home/omm/gs_bak2021
log-rotation-size = 1GB
# Retention parameters
retention-redundancy = 2
# Compression parameters
compress-algorithm = zlib
compress-level = 6[omm@prod ~]$ cat /home/omm/gs_bak2021/backups/gs_bak2021_inst/QMRIA3/backup.control
#Configuration
backup-mode = PTRACK
stream = true
compress-alg = none
compress-level = 1
from-replica = false#Compatibility
block-size = 8192
xlog-block-size = 8192
checksum-version = 0
program-version = 2.4.2
server-version = 9.2#Result backup info
timelineid = 1
start-lsn = 0/27000028
stop-lsn = 0/270002F8
start-time = '2021-01-11 16:58:51+08'
end-time = '2021-01-11 16:59:00+08'
recovery-xid = 20900
recovery-time = '2021-01-11 16:58:52+08'
expire-time = '2021-01-21 16:58:52+08'
recovery-name = 'backup QMRIA3'
data-bytes = 565410682
wal-bytes = 16777216
uncompressed-bytes = 548633410
pgdata-bytes = 592746818
status = OK
parent-backup-id = 'QMRHIA'
note = 'incr_bak2'
content-crc = 1055823518
  • 远程连接备份
## Server端用户及权限配置
postgres=# create user rep1 with sysadmin replication identified by '*****@***';  --rep1权限：sysadmin+replication
CREATE ROLE
postgres=# \du rep1List of rolesRole name |      Attributes       | Member of
-----------+-----------------------+-----------rep1      | Replication, Sysadmin | {}[omm@prod ~]$ gs_guc reload -N all -I all -h "host all all ***.***.***.***/32 sha256"    ## 放开客户端对server端的连接
[omm@prod ~]$ gs_guc reload -N all -I all -h "host replication rep1 ***.***.***.***/32 sha256"    ## 放开rep1用户对server端的replication权限
[omm@stb1 ~]$ ssh-copy-id ***.***.***.***          ## 配置SSH互信
[omm@stb1 ~]$ ssh-copy-id stb1.opengauss.com    ## 配置SSH互信## 备份客户端配置
1. SSH互信
[omm@stb1 ~]$ ssh-copy-id ***.***.***.***          ## 配置SSH互信
[omm@stb1 ~]$ ssh-copy-id prod.opengauss.com    ## 配置SSH互信2. 初始化
[omm@stb1 ~]$ mkdir -p /home/omm/gs_bak/
[omm@stb1 ~]$ gs_probackup init -B /home/omm/gs_bak
INFO: Backup catalog '/home/omm/gs_bak' successfully inited3. 添加实例
[omm@stb1 ~]$ gs_probackup add-instance -B /home/omm/gs_bak -D /gauss/data/db1 --instance='remote_prod'  \
> --remote-host=***.***.***.***  \       ## 远程Server主机
> --remote-port=22 \		      ## 远程连接端口(默认22端口)
> --remote-proto=ssh \                ## 远程连接协议(默认ssh)
> --remote-path=/gauss/app/bin \      ## 远程Server主机的gs_probackup程序所在目录
> --remote-user=omm                   ## 远程ssh连接所使用的操作系统用户
LOG: Start SSH client process, pid 1714
INFO: Instance 'remote_prod' successfully inited4. 执行远程备份
[omm@stb1 ~]$ gs_probackup backup -B /home/omm/gs_bak --instance=remote_prod -b full -D /gauss/data/db1 \
> -h ***.***.***.*** -p 26000 -d postgres -U rep1 -W *****@*** \
> --remote-host=***.***.***.*** --remote-proto=ssh --remote-port=22 --remote-user=omm --remote-path=/gauss/app/bin
INFO: Backup start, gs_probackup version: 2.4.2, instance: remote_prod, backup ID: QMSXKL, backup mode: FULL, wal mode: STREAM, remote: true, compress-algorithm: none, compress-level: 1
LOG: Backup destination is initialized
WARNING: This openGauss instance was initialized without data block checksums. gs_probackup have no way to detect data block corruption without them. Reinitialize PGDATA with option '--data-checksums'.
LOG: Start SSH client process, pid 1751
LOG: Database backup start
LOG: started streaming WAL at 0/2D000000 (timeline 1)check identify system successsend START_REPLICATION 0/2D000000 successkeepalive message is receivedkeepalive message is received
LOG: SSH process 1751 is terminated with status 0
INFO: PGDATA size: 566MB
INFO: Start transferring data files
LOG: Start SSH client process, pid 1756
LOG: Creating page header map "/home/omm/gs_bak/backups/remote_prod/QMSXKL/page_header_map"keepalive message is received
LOG: SSH process 1756 is terminated with status 0
INFO: Data files are transferred, time elapsed: 6skeepalive message is receivedkeepalive message is received
INFO: wait for pg_stop_backup()
INFO: pg_stop backup() successfully executed
LOG: stop_lsn: 0/2D0001E0
LOG: Looking for LSN 0/2D0001E0 in segment: 00000001000000000000002D
LOG: Found WAL segment: /home/omm/gs_bak/backups/remote_prod/QMSXKL/database/pg_xlog/00000001000000000000002D
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak/backups/remote_prod/QMSXKL/database/pg_xlog/00000001000000000000002D"
LOG: Found LSN: 0/2D0001E0
(null): not renaming 00000001000000000000002E, segment is not complete.
LOG: finished streaming WAL at 0/2E000140 (timeline 1)
LOG: Getting the Recovery Time from WAL
LOG: Thread [0]: Opening WAL segment "/home/omm/gs_bak/backups/remote_prod/QMSXKL/database/pg_xlog/00000001000000000000002D"
INFO: Syncing backup files to disk
INFO: Backup files are synced, time elapsed: 1s
INFO: Validating backup QMSXKL
INFO: Backup QMSXKL data files are valid
INFO: Backup QMSXKL resident size: 598MB
INFO: Backup QMSXKL completed[omm@stb1 ~]$ gs_probackup show -B /home/omm/gs_bak/BACKUP INSTANCE 'remote_prod'
======================================================================================================================================Instance     Version  ID      Recovery Time           Mode  WAL Mode  TLI  Time   Data   WAL  Zratio  Start LSN   Stop LSN    Status
======================================================================================================================================remote_prod  9.2      QMSXKL  2021-01-12 11:26:52+08  FULL  STREAM    1/0   11s  582MB  16MB    0.97  0/2D000028  0/2D0001E0  OK
[omm@stb1 ~]$ du -sh /home/omm/gs_bak/
602M    /home/omm/gs_bak/

小结

  1. 使用 gs_probackup 时，需要通过-d 指定备份时连接的数据库，否则默认会使用和当前用户同名的数据库，这样会因为这个数据库不存在而导致报错；
  2. 从备份文件看，gs_probackup 工具属于物理备份，虽然备份时需要连接特定的数据库，但是实际上备份的是整个 database cluster；
  3. 增量恢复时，gs_probackup 会验证所有所需的全备和增备文件，然后根据全备+增备文件顺序恢复，用户仅需要指定恢复目标(增备文件 ID/stop lsn 等 recovery_target)即可；
  4. 部分介绍可以参考 pg 的资料https://postgrespro.com/docs/postgrespro/9.6/app-pgprobackup个人感觉gs_probackup工具和Oracle 的 RMAN 有点相似，但在功能上、灵活程度、备份的可靠性等方面与 RMAN 还是有一些差距的，期待 openGauss 能有所完善。

openGauss 的备份工具目前已有 gs_probackup、gs_backup、gs_basebackup、gs_dump、gs_dumpall，种类较多且适用场景不同，期待内核研发的大佬们将来能对这些工具做适当调整和融合，简化使用的复杂度。

openGauss 文档有部分参数描述不够详细，这个大家可以参考 pg 的文档：Postgres Pro Standard : Documentation: 9.6: pg_probackup : Postgres Professional

本文来自互联网用户投稿，该文观点仅代表作者本人，不代表本站立场。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如若转载，请注明出处：http://www.rhkb.cn/news/288606.html

如若内容造成侵权/违法违规/事实不符，请联系长河编程网进行投诉反馈email:809451989@qq.com，一经查实，立即删除！

相关文章

Unity中如何实现草的LOD

Unity中如何实现草的LOD

1）Unity中如何实现草的LOD 2）用Compute Shader处理图像数据后在安卓机上不能正常显示渲染纹理 3）关于进游戏程序集加载的问题 4）预制件编辑模式一直在触发自动保存 这是第379篇UWA技术知识分享的推送，精选了UWA社区的热…
阅读更多...
STM32启动文件命名方式说明以及启动过程分析

STM32启动文件命名方式说明以及启动过程分析

1、启动文件的路径 cl：互联型产品，stm32f105/107系列 vl：超值型产品，stm32f100系列 xl：超高密度产品，stm32f101/103系列 flash容量大小： ld：小容量产品， 小于64KB md…
阅读更多...
利用Python进行数据可视化Plotly与Dash的应用【第157篇—数据可视化】

利用Python进行数据可视化Plotly与Dash的应用【第157篇—数据可视化】

👽发现宝藏 前些天发现了一个巨牛的人工智能学习网站，通俗易懂，风趣幽默，忍不住分享一下给大家。【点击进入巨牛的人工智能学习网站】。 利用Python进行数据可视化Plotly与Dash的应用 数据可视化是数据分析中至关重要的一环&…
阅读更多...
数字身份的革命：解锁 Web3 的身份验证技术

数字身份的革命：解锁 Web3 的身份验证技术

引言 随着数字化时代的到来，个人身份认证成为了日常生活和商业活动中不可或缺的一部分。传统的身份验证方式存在着安全性低、易伪造、不便利等问题，因此，人们迫切需要一种更安全、更便捷的身份验证技术。在这样的背景下，Web3的身…
阅读更多...
Axure中后台系统原型模板，B端页面设计实例，高保真高交互54页

Axure中后台系统原型模板，B端页面设计实例，高保真高交互54页

作品概况 页面数量：共 50 页（长期更新） 兼容版本：Axure RP 9/10，不支持低版本 应用领域：网页模板、网站后台、中台系统、B端系统 作品特色 本品为「web中后台系统页面设计实例模板」，默林原创…
阅读更多...
Linux的启动流程、模块管理与Loader

Linux的启动流程、模块管理与Loader

文章目录 Linux的启动流程BIOS、boot loader与kernel加载 内核与内核模块内核模块与依赖性：depmod查看内核模块：lsmod、modinfo内核模块的加载与删除：insmod、rmmod、modprobe内核模块的额外参数设置：/etc/modprobe.d/*.conf Linu…
阅读更多...
如何处理Flutter应用程序中的内存泄漏

如何处理Flutter应用程序中的内存泄漏

大家好，我是咕噜铁蛋！今天，我想和大家分享一下如何处理Flutter应用程序中的内存泄漏问题。在Flutter开发中，内存泄漏是一个常见且需要重点关注的问题，它可能会导致应用程序性能下降，甚至引发崩溃。因此&…
阅读更多...
PASSL代码解读[01] readme

PASSL代码解读[01] readme

介绍 PASSL 是一个基于 PaddlePaddle 的视觉库，用于使用 PaddlePaddle 进行最先进的视觉自监督学习研究。PASSL旨在加速自监督学习的研究周期：从设计一个新的自监督任务到评估所学的表征。 PASSL 主要特性： 自监督前沿算法实现 PASSL 实现了…
阅读更多...
嵌入式开发——基础电路知识

嵌入式开发——基础电路知识

1. 电路知识 1.1. 驱动能力 IC是数字逻辑芯片，其输出的是逻辑电平。逻辑电平0表示输出电压低于阈值电压，逻辑1表示输出电压高于阈值电压。负载则是被驱动的电路或元件，负载大小则指负载的电阻大小。 驱动能力主要表现在几个方面&#xff1…
阅读更多...
基于Pytorch的验证码识别模型应用

基于Pytorch的验证码识别模型应用

前言 在做OCR文字识别的时候，或多或少会接触一些验证码图片，这里收集了一些验证码图片，可以对验证码进行识别，可以识别4到6位，纯数字型、数字字母型和纯字母型的一些验证码，准确率还是相当高，需…
阅读更多...
Self-Consistency Improves Chain of Thought Reasoning in Language Models阅读笔记

Self-Consistency Improves Chain of Thought Reasoning in Language Models阅读笔记

论文链接：https://arxiv.org/pdf/2203.11171.pdf 又到了读论文的时间，内心有点疲惫。这几天还是在看CoT的文章，今天这篇是讲如何利用self-consistency（自我一致性）来改进大语言模型的思维链推理过程。什么是self-cons…
阅读更多...
设置asp.net core WebApi函数输入和返回类型中的属性名称开头大小写格式

设置asp.net core WebApi函数输入和返回类型中的属性名称开头大小写格式

以下列类型定义为例创建简单的ASP.NET Core的WebApi函数，此时输入参数和返回结果的属性名称开头默认为小写，如下图所示。 public class UserInfo { public string UserName { get; set; }public string UserSex { get; set; }public string UserP…
阅读更多...
班级综合测评管理系统的设计与实现|Springboot+ Mysql+Java+ B/S结构(可运行源码+数据库+设计文档)

班级综合测评管理系统的设计与实现|Springboot+ Mysql+Java+ B/S结构(可运行源码+数据库+设计文档)

本项目包含可运行源码数据库LW，文末可获取本项目的所有资料。 推荐阅读100套最新项目持续更新中..... 2024年计算机毕业论文（设计）学生选题参考合集推荐收藏（包含Springboot、jsp、ssmvue等技术项目合集） 目录 1. …
阅读更多...
【pytest、playwright】allure报告生成视频和图片

【pytest、playwright】allure报告生成视频和图片

目录 1、修改插件pytest_playwright 2、conftest.py配置 3、修改pytest.ini文件 4、运行case 5、注意事项 1、修改插件pytest_playwright pytest_playwright.py内容如下： # Copyright (c) Microsoft Corporation. # # Licensed under the Apache License, Ver…
阅读更多...
PL/SQL概述

PL/SQL概述

oracle从入门到总裁:​​​​​​https://blog.csdn.net/weixin_67859959/article/details/135209645 PL/SQL概述 PL/SQL(Procedural Language extension to SQL)是 Oracle 对标准 SQL语言的扩充，是专门用于各种环境下对 Oracle 数据库进行访问和开发的语言。 由…
阅读更多...
服务器呀服务器，一个虚拟专用服务器的使用教程

服务器呀服务器，一个虚拟专用服务器的使用教程

目前刚接触服务器这一块的学习，这里记录一下解如何获取自己的第一台虚拟云服务器，给刚入行服务器开发的小伙伴做一个参考。 具体的步骤如下： 一、服务器的注册和获取 1、打开bwg88服务器平台地址：点击进入 https://bwh88.net/a…
阅读更多...
基于云计算的前端资源管理系统的设计与实现

基于云计算的前端资源管理系统的设计与实现

hello宝子们...我们是艾斯视觉擅长ui设计和前端开发10年经验！希望我的分享能帮助到您！如需帮助可以评论关注私信我们一起探讨！致敬感谢感恩！ 随着互联网的快速发展，前端资源管理成为了一个重要的课题。本文旨在设计并实…
阅读更多...
新书速递——《可解释AI实战（PyTorch版）》

新书速递——《可解释AI实战（PyTorch版）》

本书旨在帮助你实施最新的可解释AI技术，以构建公平且可解释的AI系统。可解释AI是当今AI研究中的热门话题，但只有少数资源和指南涵盖了所有重要技术，这些技术对实践者来说非常有价值。本书旨在填补这一空白。 本书读者对象 本书既适合那些有兴…
阅读更多...
BUG定位---一起学习吧之测试

BUG定位---一起学习吧之测试

判断一个BUG是前端还是后端的，通常需要根据BUG的具体表现、发生的环境以及相关的技术栈来进行分析。以下是一些常用的判断方法： 错误发生的位置： 如果BUG涉及的是页面的布局、样式、交互效果等，那么很可能是前端的BUG。如果BUG与…
阅读更多...
LNMP架构之mysql数据库实战

LNMP架构之mysql数据库实战

mysql安装 到官网www.mysql.com下载源码版本 实验室使用5.7.40版本 tar xf mysql-boost-5.7.40.tar.gz #解压 cd mysql-boost-5.7.40/ yum install -y cmake gcc-c bison #安装依赖性 cmake -DCMAKE_INSTALL_PREFIX/usr/local/mysql -DMYSQL_DATADIR/data/mysql -DMYSQL_…
阅读更多...
最新文章
推荐文章

Copyright @ 2022~2023