From 88a6a4f2a1c8889693b6f69f1e49c9d23ebc4cc4 Mon Sep 17 00:00:00 2001 From: Alina Derkach Date: Wed, 30 Oct 2024 15:25:12 +0200 Subject: [PATCH 1/2] PXB-3254 Document reduction in locks modified: docs/lock-options.md new file: docs/reduction-in-locks.md modified: docs/xtrabackup-option-reference.md modified: mkdocs-base.yml --- docs/how-xtrabackup-works.md | 9 +-- docs/reduction-in-locks.md | 118 ++++++++++++++++++++++++++++ docs/xtrabackup-option-reference.md | 33 ++++++-- mkdocs-base.yml | 3 +- snippets/pro-build-announcement.md | 3 + 5 files changed, 153 insertions(+), 13 deletions(-) create mode 100644 docs/reduction-in-locks.md create mode 100644 snippets/pro-build-announcement.md diff --git a/docs/how-xtrabackup-works.md b/docs/how-xtrabackup-works.md index a75ab4da..4651d9e5 100644 --- a/docs/how-xtrabackup-works.md +++ b/docs/how-xtrabackup-works.md @@ -16,12 +16,7 @@ The `--register-redo-log-consumer` parameter is disabled by default. When enable Percona XtraBackup remembers the LSN when it starts, and then copies the data files. The operation takes time, and the files may change, then LSN reflects the state of the database at different points in time. Percona XtraBackup also runs a background process that watches the transaction log files, and copies any changes. Percona XtraBackup does this continually. The transaction logs are written in a round-robin fashion, and can be reused. -Percona XtraBackup uses [Backup locks] - -where available as a lightweight alternative to `FLUSH TABLES WITH READ -LOCK`. MySQL {{vers}} allows -acquiring an instance level backup lock via the `LOCK INSTANCE FOR BACKUP` -statement. +Percona XtraBackup uses [Backup locks] where available as a lightweight alternative to `FLUSH TABLES WITH READ LOCK`. MySQL {{vers}} allows acquiring an instance level backup lock via the `LOCK INSTANCE FOR BACKUP` statement. Locking is only done for MyISAM and other non-InnoDB tables after Percona XtraBackup finishes backing up all InnoDB/XtraDB data and @@ -87,7 +82,7 @@ to do operations such as streaming and incremental backups with various combinations of copying the data files, copying the log files, and applying the logs to the data. -## Restoring a backup +## Restore a backup To restore a backup with xtrabackup you can use the `--copy-back` or `--move-back` options. diff --git a/docs/reduction-in-locks.md b/docs/reduction-in-locks.md new file mode 100644 index 00000000..880cd6d0 --- /dev/null +++ b/docs/reduction-in-locks.md @@ -0,0 +1,118 @@ +# Reduced backup lock time + +--8<--- "pro-build-announcement.md" + +!!! important + + The `--lock-ddl=REDUCED` option is a [tech preview](./glossary.md#tech-preview). Before using this option in production, we recommend that you test restoring production from physical backups in your environment, and also use the alternative backup method for redundancy. + +[Percona XtraBackup 8.4.0-2](./release-notes/8.4.0-2.md) adds the [`--lock-ddl=REDUCED`](./xtrabackup-option-reference.md#lock-ddl) option to reduce the time the server remains locked by `xtrabackup` during full and incremental backups. Now, you can execute `Data Definition Language` (DDL) operations while the backup is in progress. + +### Benefits + +The `--lock-ddl=REDUCED` option features are as follows: + +* Acquire and release the backup lock quickly: The backup process minimizes the time it holds the backup lock, allowing for concurrent DDL operations. + + ??? note "Comparing the backup duration with --lock-ddl=REDUCED" + + While we strive to provide accurate performance benchmarks, real-world results may vary depending on your hardware and software configurations. + + The following tables illustrate the approximate time differences between `--lock-ddl=ON` and `--lock-ddl=REDUCED` for local and cloud backups (Amazon S3). The "Improvement (X times)" column shows how many times faster the backup is when using `--lock-ddl=REDUCED` compared to `--lock-ddl=ON`. + + === "Local backup" + + | Data size in gigabytes | Backup duration with --lock-ddl=ON in seconds| Backup duration with --lock-ddl=REDUCED in seconds| Improvement (X times) | + |-----------------------------|---------------|--------------------|--------------------------| + | 50 GB | 460.2 sec | 2.169 sec | 212.17 | + | 100 GB | 901.8 sec | 1.305 sec | 691.03 | + | 200 GB | 1820.4 sec | 1.347 sec | 1351.45 | + | 400 GB | 3580.2 sec | 1.24 sec | 2887.26 | + | 500 GB | 5436 sec | 1.264 sec | 4300.63 | + + === "Cloud (Amazon s3) backup" + + | Data size in gigabytes | Backup duration with --lock-ddl=ON in seconds | Backup duration with --lock-ddl=REDUCED in seconds | Improvement (X times) | + |----------------------------|---------------|--------------------|--------------------------| + | 50 GB | 469.8 sec | 3.859 sec | 121.74 | + | 100 GB | 927.6 sec | 4.072 sec | 227.80 | + | 200 GB | 1851.6 sec | 4.408 sec | 420.05 | + | 400 GB | 3888 sec | 3.948 sec | 984.80 | + | 500 GB | 4896 sec | 4.065 sec | 1204.43 | + + Using `--lock-ddl=REDUCED` leads to a dramatic reduction in backup time compared to `--lock-ddl=ON`, especially with larger data sizes. + +* Track changes with redo logs: Redo logs record all file-level changes, ensuring data consistency during the backup process. +* Handle DDL operations: The backup process generates metadata files to account for any DDL operations that occur while the backup is in progress. +* Ensure consistency: The `--prepare` step processes generated metadata files and uses redo and undo logs to create a consistent database state. + +### Limitations + +* Certain DDL operations consume significant resources, and xtrabackup simultaneously requires I/O to copy and read files. This requirement can increase resource demand from DDL operations during the backup process. + +* The master key rotation with the `ALTER INSTANCE ROTATE INNODB MASTER KEY` is prohibited while backup is in progress. + +* Backups may be larger because new `DDL` operations are executed concurrently on the server, and the files generated by the server are included in the backup. Additionally, certain `DDL` operations, such as `ADD INDEX` or encryption changes to existing data files, will cause the data files to be recopied, increasing the backup size. + +* The number of open file handles in your operating system should be configured to match the number of files in the server data directory. + +* Taking a backup using [page tracking](./page-tracking.md) is not supported with `--lock-ddl=REDUCED` option. + +## How the `--lock-ddl=REDUCED` option works + +Percona Server for MySQL implemented `LOCK TABLES FOR BACKUP` to block `DDL` operations on a server. `xtrabackup` utilizes this lock during the backup process to ensure that any `DDL` operation does not corrupt the backup. With MySQL-compatible servers `xtrabackup` utilizes `LOCK INSTANCE FOR BACKUP` to lock the server during the backup. The lock does not affect the `Data Manipulation Language` (DML) operations. By default, `xtrabackup` enables the [`--lock-ddl=ON`](./xtrabackup-option-reference.md#lock-ddl) option to apply the backup lock at the start of the backup. + +Taking backup with the `--lock-ddl=REDUCED` option includes the following phases: + +### Phase 1: Operations performed without the lock + +!!! warning + + Do not execute the `ALTER INSTANCE ROTATE MASTER KEY` command during `Phase 1` if there are encrypted tables. This prevents backup failures. + +The following operations are performed without the lock: +{ .power-number } + +1. Parse and copy the redo logs from the checkpoint up to the current `LSN`, and start tracking new file operations. +2. Start the redo log thread to copy the redo logs. +3. Track file operations by parsing the `MLOG_FILE_*` records from the redo log. These records help track changes in the files being backed up to ensure consistency. +4. Copy the `.ibd` files. + +### Phase 2: Operations performed under the lock + +The following operations are performed under the lock: +{ .power-number } + +1. Take the backup lock on the server to prevent new DDL operations, such as creating or altering tables. +2. Copy `non-InnoDB` files. +3. Check the file operations that were tracked and recopy the tablespaces. +4. Create additional `metadata` files to perform the required actions (deletions or renames) on the already copied files. This approach ensures that the backup remains consistent and accurate without disrupting the streaming process. This is particularly important for streaming backups. +5. Gather a sync point from all engines (InnoDB LSN, binlog, GTID, etc.) by querying the `log_status`. +6. Stop the redo thread once it copies at least up to sync point at step 4. +7. Release the backup lock on the server. + +??? note "The list of metadata files" + + | File | Description | + |--------------|------------------------------------------------------------------------------------------------------------------| + | `.new` | For new files that are created while the backup in progress and for files that need to be recopied due to encryption or `ADD INDEX` operations. | + | `.del` | For deleted files. For a file that has been copied by backup but deleted while the backup is in progress, create a `space_id.del` file. | + | `.ren` | For a file that has been copied but renamed while the backup is in progress, create a space_id.ren file. This file should contain the new file name. | + | `.crpt` | For files that cannot be fully copied due to encryption changes. These files will be recopied, and a `.crpt` file will be created for the existing file. | + | `.new.meta` | Similar to `.new`, but for incremental backups. | + | `.new.delta` | Similar to `.new`, but for incremental backups. Create `t1.ibd.meta` and `t1.ibd.delta` instead of `t1.ibd`. | + +### Phase 3. Handling new metadata files in `--prepare` + +The following operations are performed to ensure data remains intact and consistent: +{ .power-number } + +1. Remove the `.crpt` files that match the name after stripping the extension. This step is crucial before the `IBD` scan because these files are incomplete (they can be zero size). +2. Perform a scan to create a mapping of `space_id` to file names. +3. Delete the file matching the `space_id` using `space_id.del`. In case of incremental backups, also delete the corresponding `.new.meta` and `.new.delta` files. +4. Rename the file matching the `space_id` to the name specified in the file using `space_id.ren`. +5. Replace the file that matches the name without the `.new` extension using the `.new` extension. +6. Replace the `meta` and `delta` files that match the name without the `.new` in the name using `.new.meta` and `.new.delta`. + +After `Phase 3` is completed, the regular [crash recovery](./how-xtrabackup-works.md) starts. + diff --git a/docs/xtrabackup-option-reference.md b/docs/xtrabackup-option-reference.md index afd45e55..2a48d284 100644 --- a/docs/xtrabackup-option-reference.md +++ b/docs/xtrabackup-option-reference.md @@ -27,7 +27,6 @@ $ xtrabackup --prepare --target-dir=/data/backup/mysql/ For all modes, the default options are read from the xtrabackup and mysqld configuration groups from the following files in the given order: - 1. `/etc/my.cnf` @@ -598,7 +597,32 @@ Usage: `--lock-ddl` Enabled by default to ensure that any DDL event does not corrupt the backup. Any DML events continue to occur. A DDL lock protects table and view definitions. -If the option is disabled, a backup continues while concurrent DDL events are executed. These backups are invalid and fail in the Prepare step. +The available values are the following: + +|Option| Description| +|-------|-----------| +|`--lock-ddl=ON`| The backup lock is enabled and is taken at the beginning of the backup.| +|`--lock-ddl=OFF`|The backup lock is not taken.| +|`--lock-ddl=REDUCED`|This option value is a [tech preview](./glossary.md#tech-preview).

The option value has been added in [Percona XtraBackup 8.4.0-2](./release-notes/8.4.0-2.md) to reduce the time the instance remains under backup lock. The backup lock is taken after copying the `.ibd` files and before copying the `non-InnoDB` files.| + +With the `--lock-ddl=ON` option the backup process is as follows: +{ .power-number } + +1. Set a backup lock on the server to prevent any DDL operations, such as creating or altering tables. + +2. Parse and copy the redo logs from the checkpoint up to the current `Log Sequence Number` (LSN), and start tracking new file operations. + +3. Identify and copy the `.ibd` files. + +4. Copy `non-InnoDB` files. + +5. Gather a sync point from all engines (InnoDB LSN, binlog, Global Transaction Identifier (GTID), etc.) by querying the log status. + +6. Stop the redo thread once it copies at least up to sync point. + +7. Release the backup lock on the server. + +If the backup lock is disabled with the [`--lock-ddl=OFF`](./xtrabackup-option-reference.md/#lock-ddl) option, a backup continues while concurrent DDL operations are executed. These backups may be invalid and may fail at either the `backup` or the `--prepare` step. Use a [safe-slave-backup](#safe-slave-backup) option to stop a SQL replica thread before copying the InnoDB files. @@ -606,10 +630,9 @@ Use a [safe-slave-backup](#safe-slave-backup) option to stop a SQL replica threa Usage: `--lock-ddl-per-table` -Deprecated. Use the [`–lock-ddl`](#lock-ddl) option instead +Deprecated in Percona XtraBackup 8.0. Use the [`–lock-ddl`](#lock-ddl) option instead -Lock DDL for each table before xtrabackup starts to copy -it and until the backup is completed. +Lock DDL for each table before xtrabackup starts to copy it and until the backup is completed. ### lock-ddl-timeout diff --git a/mkdocs-base.yml b/mkdocs-base.yml index c7af762c..a94f8d44 100644 --- a/mkdocs-base.yml +++ b/mkdocs-base.yml @@ -164,9 +164,10 @@ nav: - Restore features: - point-in-time-recovery.md - restore-individual-tables.md + - Pro build features: + - reduction-in-locks.md - flush-tables-with-read-lock.md - log-enhancements.md - - lock-options.md - smart-memory-estimation.md - working-with-binary-logs.md - Binaries: diff --git a/snippets/pro-build-announcement.md b/snippets/pro-build-announcement.md new file mode 100644 index 00000000..e9fb23e0 --- /dev/null +++ b/snippets/pro-build-announcement.md @@ -0,0 +1,3 @@ +Percona XtraBackup Pro includes the [capabilities](pxb-pro.md#capabilities) that are typically requested by large enterprises. Percona XtraBackup Pro contains packages created and tested by Percona. These packages are supported only for Percona Customers with a subscription. + +[Become a Percona Customer](https://www.percona.com/about/contact){.md-button} \ No newline at end of file From 50c9e22896a31b21dae2462dec25a02f8bf6b0cb Mon Sep 17 00:00:00 2001 From: Alina Derkach Date: Fri, 13 Dec 2024 17:03:16 +0100 Subject: [PATCH 2/2] Update reduction-in-locks.md --- docs/reduction-in-locks.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/reduction-in-locks.md b/docs/reduction-in-locks.md index 880cd6d0..1ccbc15b 100644 --- a/docs/reduction-in-locks.md +++ b/docs/reduction-in-locks.md @@ -18,7 +18,7 @@ The `--lock-ddl=REDUCED` option features are as follows: While we strive to provide accurate performance benchmarks, real-world results may vary depending on your hardware and software configurations. - The following tables illustrate the approximate time differences between `--lock-ddl=ON` and `--lock-ddl=REDUCED` for local and cloud backups (Amazon S3). The "Improvement (X times)" column shows how many times faster the backup is when using `--lock-ddl=REDUCED` compared to `--lock-ddl=ON`. + The following tables illustrate the approximate time differences between `--lock-ddl=ON` and `--lock-ddl=REDUCED` for local and cloud backups (Amazon S3). The `Improvement (X times)` column shows how many times less the server is locked by xtrabackup using `--lock-ddl=REDUCED` compared to `--lock-ddl=ON`. === "Local backup"