This repository includes backup and recovery utilities for GitHub Enterprise.
- Features
- Requirements
- Getting started
- Migrating from GitHub Enterprise v11.10.34x to v2.0
- Using the backup and restore commands
- Scheduling backups
- Backup snapshot file structure
- How does backup utilities differ from a High Availability replica?
- Support
The backup utilities implement a number of advanced capabilities for backup hosts, built on top of the backup and restore features already included in GitHub Enterprise.
- Complete GitHub Enterprise backup and recovery system via two simple utilities:
ghe-backupandghe-restore. - Online backups. The GitHub appliance need not be put in maintenance mode for the duration of the backup run.
- Incremental backup of Git repository data. Only changes since the last snapshot are transferred, leading to faster backup runs and lower network bandwidth and machine utilization.
- Efficient snapshot storage. Only data added since the previous snapshot consumes new space on the backup host.
- Multiple backup snapshots with configurable retention periods.
- Backup commands run under the lowest CPU/IO priority on the GitHub appliance, reducing performance impact while backups are in progress.
- Runs under most Linux/Unix environments.
- MIT licensed, open source software maintained by GitHub, Inc.
The backup utilities should be run on a host dedicated to long-term permanent storage and must have network connectivity with the GitHub Enterprise appliance.
Backup host software requirements are modest: Linux or other modern Unix operating system with bash, git, and rsync v2.6.4 or newer.
The backup host must be able to establish network connections outbound to the GitHub appliance over SSH. TCP port 122 is used to backup GitHub Enterprise 2.0 or newer instances, and TCP port 22 is used for older versions (11.10.34X).
Storage requirements vary based on current Git repository disk usage and growth patterns of the GitHub appliance. We recommend allocating at least 5x the amount of storage allocated to the primary GitHub appliance for historical snapshots and growth over time.
The backup utilities use hard links to store data efficiently, so the backup snapshots must be written to a filesystem with support for hard links.
The backup utilities are fully supported under GitHub Enterprise 2.0 or greater.
The previous release series (11.10.34x) is also supported but must meet minimum version requirements. For online and incremental backup support, the GitHub Enterprise instance must be running version 11.10.342 or above.
Earlier versions are supported, but online and incremental backups are not supported. We strongly recommend upgrading to the latest release if you're running a version prior to 11.10.342. Visit enterprise.github.com to download the most recent GitHub Enterprise version.
Note: You can restore a snapshot that's at most two feature releases behind the restore target's version of GitHub Enterprise. For example, to restore a snapshot of GitHub Enterprise 2.4, the target GitHub Enterprise appliance must be running GitHub Enterprise 2.5.x or 2.6.x. You can't restore a snapshot from 2.4 to 2.7, because that's three releases ahead.
-
Download the latest release version and extract or clone the repository using Git:
git clone -b stable https://github.com/github/backup-utils.git -
Copy the
backup.config-examplefile tobackup.configand modify as necessary. TheGHE_HOSTNAMEvalue must be set to the GitHub Enterprise host name. Additional options are available and documented in the configuration file but none are required for basic backup functionality.-
backup-utils will attempt to load the backup configuration from the following locations, in this order:
$GHE_BACKUP_CONFIG (User configurable environment variable) $GHE_BACKUP_ROOT/backup.config (Root directory of backup-utils install) $HOME/.github-backup-utils/backup.config /etc/github-backup-utils/backup.config -
In a clustering environment, the
GHE_EXTRA_SSH_OPTSkey must be configured with the-i <abs path to private key>SSH option.
-
-
Add the backup host's SSH key to the GitHub appliance as an Authorized SSH key. See Adding an SSH key for shell access for instructions.
-
Run
bin/ghe-host-checkto verify SSH connectivity with the GitHub appliance. -
Run
bin/ghe-backupto perform an initial full backup.
If you are migrating from GitHub Enterprise version 11.10.34x to 2.0 or 2.1 (note, migrations to versions greater than 2.1 are not officially supported), please see the Migrating from GitHub Enterprise v11.10.34x documentation in the GitHub Enterprise System Administrator's Guide. It includes important information on using the backup utilities to migrate data from your v11.10.34x instance to v2.0 or v2.1.
After the initial backup, use the following commands:
- The
ghe-backupcommand creates incremental snapshots of repository data, along with full snapshots of all other pertinent data stores. - The
ghe-restorecommand restores snapshots to the same or separate GitHub Enterprise appliance. You must add the backup host's SSH key to the target GitHub Enterprise appliance before using this command.
The following assumes that GHE_HOSTNAME is set to "github.example.com" in
backup.config.
Creating a backup snapshot:
$ ghe-backup
Starting backup of github.example.com in snapshot 20140727T224148
Connect github.example.com OK (v11.10.343)
Backing up GitHub settings ...
Backing up SSH authorized keys ...
Backing up SSH host keys ...
Backing up MySQL database ...
Backing up Redis database ...
Backing up Git repositories ...
Backing up GitHub Pages ...
Backing up Elasticsearch indices ...
Completed backup of github.example.com in snapshot 20140727T224148 at 23:01:58
Restoring from last successful snapshot to a newly provisioned GitHub Enterprise appliance at IP "5.5.5.5":
$ ghe-restore 5.5.5.5
Starting rsync restore of 5.5.5.5 from snapshot 20140727T224148
Connect 5.5.5.5 OK (v11.10.343)
Enabling maintenance mode on 5.5.5.5 ...
Restoring Git repositories ...
Restoring GitHub Pages ...
Restoring MySQL database ...
Restoring Redis database ...
Restoring SSH authorized keys ...
Restoring Elasticsearch indices ...
Restoring SSH host keys ...
Completed restore of 5.5.5.5 from snapshot 20140817T174152
Visit https://5.5.5.5/setup/settings to configure the recovered appliance.
A different backup snapshot may be selected by passing the -s argument and the
datestamp-named directory from the backup location.
The ghe-backup and ghe-restore commands also have a verbose output mode
(-v) that lists files as they're being transferred. It's often useful to
enable when output is logged to a file.
When restoring to an already configured GHE instance, settings, certificate, and license data
are not restored to prevent overwriting manual configuration on the restore
host. This behavior can be overridden by passing the -c argument to ghe-restore,
forcing settings, certificate, and license data to be overwritten with the backup copy's data.
Regular backups should be scheduled using cron(8) or similar command
scheduling service on the backup host. The backup frequency will dictate the
worst case recovery point objective (RPO) in your backup plan. We recommend the
following:
- Hourly backups for GitHub Enterprise versions 11.10.342 or greater (due to improved online and incremental backup support)
- Daily backups for versions prior to 11.10.342.
Note: the time required to do full offline backups of large datasets under GitHub Enterprise versions prior to 11.10.342 may prohibit the use of daily backups. We strongly recommend upgrading to 11.10.342 or greater in that case.
The following examples assume the backup utilities are installed under
/opt/backup-utils. The crontab entry should be made under the same user that
manual backup/recovery commands will be issued under and must have write access
to the configured GHE_DATA_DIR directory.
Note that the GHE_NUM_SNAPSHOTS option in backup.config should be tuned
based on the frequency of backups. The ten most recent snapshots are retained by
default. The number should be adjusted based on backup frequency and available
storage.
To schedule hourly backup snapshots with verbose informational output written to a log file and errors generating an email:
MAILTO=admin@example.com
0 * * * * /opt/backup-utils/bin/ghe-backup -v 1>>/opt/backup-utils/backup.log 2>&1
To schedule nightly backup snapshots instead, use:
MAILTO=admin@example.com
0 0 * * * /opt/backup-utils/bin/ghe-backup -v 1>>/opt/backup-utils/backup.log 2>&1
Backup snapshots are stored in rotating increment directories named after the date and time the snapshot was taken. Each snapshot directory contains a full backup snapshot of all relevant data stores. Repository, Search, and Pages data is stored efficiently via hard links.
Please note Symlinks must be maintained when archiving backup snapshots. Dereferencing or excluding symlinks, or storing the snapshot contents on a filesystem which does not support symlinks will result in operational problems when the data is restored.
The following example shows a snapshot file hierarchy for hourly frequency.
There are five snapshot directories, with the current symlink pointing to the
most recent successful snapshot:
./data
|- 20140724T010000
|- 20140725T010000
|- 20140726T010000
|- 20140727T010000
|- 20140728T010000
|- authorized-keys.json
|- elasticsearch/
|- enterprise.ghl
|- mysql.sql.gz
|- pages/
|- redis.rdb
|- repositories/
|- settings.json
|- ssh-host-keys.tar
|- strategy
|- version
|- current -> 20140728T010000
Note: the GHE_DATA_DIR variable set in backup.config can be used to change
the disk location where snapshots are written.
It is recommended that both backup utilities and an High Availability replica are used as part of a GitHub Enterprise deployment but they serve different roles.
The High Availability replica is a fully redundant secondary GitHub Enterprise instance, kept in sync with the primary instance via replication of all major datastores. This active/passive cluster configuration is designed to minimize service disruption in the event of hardware failure or major network outage affecting the primary instance. Because some forms of data corruption or loss may be replicated immediately from primary to replica, it is not a replacement for the backup utilities as part of your disaster recovery plan.
Backup utilities are a disaster recovery tool. This tool takes date-stamped snapshots of all major datastores. These snapshots are used to restore an instance to a prior state or set up a new instance without having another always-on GitHub Enterprise instance (like the High Availability replica).
If you find a bug or would like to request a feature in backup-utils, please open an issue or pull request on this repository. If you have a question related to your specific GitHub Enterprise setup or would like assistance with backup site setup or recovery, please contact our Enterprise support team instead.