Skip to content

Percona Backup for MongoDB 1.7.0

Date

April 18, 2022

Installation

Installing Percona Backup for MongoDB

Percona Backup for MongoDB is a distributed, low-impact solution for consistent backups of MongoDB sharded clusters and replica sets. This is a tool for creating consistent backups across a MongoDB sharded cluster (or a single replica set), and for restoring those backups to a specific point in time. The project was inspired by (and intends to replace) the Percona-Lab/mongodb_consistent_backup tool.

Important

To make physical backups and restores, the pbm-agent must have the read / write access to the dataDir. If you use the filesystem-based backup storage, the pbm-agent must also have the read / write access to the backup directory. Therefore, starting from version 1.7.0, the user running the pbm-agent is changed from pbm to mongod in Percona Backup for MongoDB packages.

To upgrade Percona Backup for MongoDB to version 1.7.0, do the following:

  1. Stop the pbm-agent process

  2. Upgrade new version packages

  3. Reload the systemd process to update the unit file with the following command:

    $ sudo  systemctl daemon-reload

  1. If you have a filesystem-based backup storage, grant read / write permissions to the backup directory to the mongod user.

  2. Restart the pbm-agent process.

If MongoDB runs under a different user than mongod (the default configuration for Percona Server for MongoDB), use the same user to run the pbm-agent. For filesystem-based storage, grant the read / write permissions to the backup directory for this user.

Release highlights

  • Support for physical backups in Percona Server for MongoDB starting from versions 4.2.15-16 and 4.4.6-8 and higher. Physical backups drastically speed up backup and restore performance for large databases (several terabytes). This is a technical preview feature.

  • Oplog replay from the arbitrary start time. This reduces Recovery Point Objective (RPO) when database is recovered from physical or storage-level backups.

  • Ability to configure compression method and level for Point-in-Time Recovery chunks and compression level for backups.

  • Ability to configure the number of S3 multipart upload chunks to comply with various S3-compatible storage provider requirements.

  • Ability to configure the number of upload retries. This facilitates data upload in case of unstable network connection.

New Features

  • PBM-734: Add the config option to set debug log levels for S3 requests

  • PBM-805: Implement physical backups to improve performance for large databases

  • PBM-742: Add the ability to replay oplog from arbitrary start time. This reduces Recovery Point Objective (RPO) when database is recovered from physical backups.

Improvements

  • PBM-680: Skip TLS verification for object storage. This can be useful for private object storage with self-signed certificates.

  • PBM-770: Support configurable compression method / level for Point-in-Time Recovery chunks (Thanks to Damiano Albani for reporting this issue and contributing to it)

  • PBM-764: Support Zstandard compression format (Thanks to Damiano Albani for reporting this issue and contributing to it)

  • PBM-750: Make max number of S3 upload parts configurable (Thanks to Damiano Albani for reporting this issue and contributing to it)

  • PBM-777: Expand / fix the configuration API to support compression method for Point-in-Time Recovery chunks (Thanks to Damiano Albani for reporting and contributing to this issue)

  • PBM-756: Add the ability to configure logging levels for S3 requests to debug issues with object storage (Thanks to Damiano Albani for reporting this issue and contributing to it)

  • PBM-577: It is now possible to choose an S3 storage class for granular control over various S3 tiers (Thanks to Damiano Albani for the contribution)

Bugs Fixed

  • PBM-721: Fixed a bug where an upload of the backup to S3-storage was failing due to unstable network connection. Percona Backup for MongoDB can now be configured to retry the upload with flexible timeouts.

  • PBM-773: Check distributed transactions on all participating shards to avoid commit timestamp inconsistency upon restore