刘凡 9ff4d1d109 add S3,archive,truncate | 2 years ago | |
---|---|---|
.. | ||
.github | 2 years ago | |
10 | 2 years ago | |
Makefile | 2 years ago | |
README.md | 2 years ago | |
update.py | 2 years ago |
Docker image that periodically dumps a Postgres database, and uploads it to an Amazon S3 bucket.
CRON_SCHEDULE
: The time schedule part of a crontab file (e.g: 15 3 * * *
for every night 03:15)DB_HOST
: Postgres hostnameDB_PASS
: Postgres passwordDB_USER
: Postgres usernameDB_NAME
: Name of databaseS3_PATH
: Amazon S3 path in the format: s3://bucket-name/some/pathAWS_ACCESS_KEY_ID
AWS_SECRET_ACCESS_KEY
AWS_DEFAULT_REGION
S3_STORAGE_CLASS
: Specify storage class for the uploaded object, defaults to STANDARD_IA
.S3_EXTRA_OPTIONS
: Specify additional options for S3, e.g. --endpoint=
for using custom S3 provider.DB_USE_ENV
: Inject postgres environment variables from the environment. Ignores DB_HOST
, DB_PASS
, DB_USER
and DB_NAME
. Can be used to specify advanced connections, e.g. using mTLS connection.
Example of --env-file=.env
for container:
DB_USE_ENV=True
PGSSLMODE=verify-full
PGSSLROOTCERT=/path/ca.crt
PGSSLCERT=<path>/user.crt
PGSSLKEY=<path>/user.key
PGHOSTADDR=1.2.3.4
PGHOST=db.domain.com
PGUSER=myuser
PGDATABASE=mydb
MAIL_TO
: If MAIL_TO
and MAIL_FROM
is specified, an e-mail will be sent, using Amazon SES, every time a backup is taken
MAIL_FROM
WEBHOOK
: If specified, an HTTP request will be sent to this URL
WEBHOOK_METHOD
: By default the webhook's HTTP method is GET, but can be changed using this variable
WEBHOOK_CURL_OPTIONS
: Add additional headers or other option to curl command calling the webhook. E.g. -H 'Content-type: application/json'
WEBHOOK_DATA
: Add a body to the webhook being called, unless changed it implies that POST
method is used. E.g. {"text":"Backup completed at %(date)s %(time)s!"}
KEEP_BACKUP_DAYS
: The number of days to keep backups for when pruning old backups. Defaults to 7
.
FILENAME
: String that is passed into strftime()
and used as the backup dump's filename. Defaults to $DB_NAME_%Y-%m-%d
.
Text in WEBHOOK_DATA
is interpolated with variabels %(my_var)s
date
: Date in yyyy-mm-dd formattime
: Date in hh:mm:ss formatduration
: Number of seconds take to backupfilename
: Name of the file uploaded to S3size
: Size of the backup file with suitable suffix, like MB, GB, ...Set WEBHOOK
and WEBHOOK_
accodringly:
WEBHOOK=https://hooks.slack.com/services/.../.../...
WEBHOOK_METHOD=POST
WEBHOOK_CURL_OPTIONS=-H 'Content-type: application/json'
WEBHOOK_DATA={"text":":white_check_mark: Backup completed at %(date)s %(time)s\nDuration: %(duration)s seconds\nUpload: %(filename)s: %(size)s"}
/data/backups
- The database is dumped in into this directoryThis image can also be run as a one off task to restore one of the backups.
To do this, we run the container with the command: python -u /backup/restore.py [S3-filename]
(S3-filename
should only be the name of the file, the directory is set through the S3_PATH
env variable).
The following environment variables are required:
DB_HOST
: Postgres hostnameDB_PASS
: Postgres passwordDB_USER
: Postgres usernameDB_NAME
: Name of database to import intoS3_PATH
: Amazon S3 directory path in the format: s3://bucket-name/some/pathAWS_ACCESS_KEY_ID
AWS_SECRET_ACCESS_KEY
AWS_DEFAULT_REGION
S3_EXTRA_OPTIONS
: Specify additional options for S3, e.g. --endpoint=
for using custom S3 provider.DB_USE_ENV
: See Optional environment variables above.To run a one off backup job, e.g. to test that it works when setting it up for the first time, simply start
the container with the docker run command set to python -u /backup/backup.py
(as well as all the required environment
variables set).
This image uses the alpine version(s) of the official postgres image as base image.
The following docker tags are available for this image, and they are based on the corresponding official postgres alpine image:
13
, latest
12
11
10