@cloudant/couchbackup/README.md

# CouchBackup

[![npm (scoped)](https://img.shields.io/npm/v/@cloudant/couchbackup.svg?colorB=0000ff)](https://www.npmjs.com/package/@cloudant/couchbackup)
[![npm (scoped with tag)](https://img.shields.io/npm/v/@cloudant/couchbackup/snapshot.svg?colorB=666699)](https://www.npmjs.com/package/@cloudant/couchbackup)
[![Build Status](https://travis-ci.org/cloudant/couchbackup.svg?branch=master)](https://travis-ci.org/cloudant/couchbackup)
[![Greenkeeper badge](https://badges.greenkeeper.io/cloudant/couchbackup.svg)](https://greenkeeper.io/)

```
 _____                  _    ______            _
/  __ \                | |   | ___ \          | |
| /  \/ ___  _   _  ___| |__ | |_/ / __ _  ___| | ___   _ _ __
| |    / _ \| | | |/ __| '_ \| ___ \/ _` |/ __| |/ / | | | '_ \
| \__/\ (_) | |_| | (__| | | | |_/ / (_| | (__|   <| |_| | |_) |
 \____/\___/ \__,_|\___|_| |_\____/ \__,_|\___|_|\_\\__,_| .__/
                                                         | |
                                                         |_|
```

CouchBackup is a command-line utility that allows a Cloudant or CouchDB database to be backed up to a text file.
It comes with a companion command-line utility that can restore the backed up data.

## Limitations

Couchbackup has some restrictions in the data it's able to backup:

* **couchbackup does not do CouchDB replication as such, it simply streams through a database's `_changes` feed, and uses `POST /db/_bulk_get` to fetch the documents, storing the documents it finds on disk.**
* **couchbackup does not support backing up or restoring databases containing documents with attachments. It is recommended to store attachments directly in an object store. DO NOT USE THIS TOOL FOR DATABASES CONTAINING ATTACHMENTS.** [Note](#note-on-attachments)

## Installation

To install the latest released version use npm:

```sh
npm install -g @cloudant/couchbackup
```

### Requirements
* The minimum required Node.js version is 12.
* The minimum required CouchDB version is 2.0.0.

### Snapshots

The latest builds of master are published to npm with the `snapshot` tag. Use the `snapshot` tag if you want to experiment with an unreleased fix or new function, but please note that snapshot versions are **unsupported**.

## Usage

Either environment variables or command-line options can be used to specify the URL of the CouchDB or Cloudant instance, and the database to work with.

### The URL

To define the URL of the CouchDB instance set the `COUCH_URL` environment variable:

```sh
export COUCH_URL=http://localhost:5984
```
or

```sh
export COUCH_URL=https://myusername:mypassword@myhost.cloudant.com
```

Alternatively we can use the `--url` command-line parameter.

When passing credentials in the user information subcomponent of the URL
they must be [percent encoded](https://tools.ietf.org/html/rfc3986#section-3.2.1).
Specifically, within either the username or password, the characters `: / ? # [ ] @ %`
_MUST_ be precent-encoded, other characters _MAY_ be percent encoded.

For example, for the username `user123` and password `colon:at@321`:
```
https://user123:colon%3aat%40321@localhost:5984
```

Note that additional care must be taken to escape shell reserved characters when
setting the environment variable or command-line parameter.

### The Database name

To define the name of the database to backup or restore, set the `COUCH_DATABASE` environment variable:

```sh
export COUCH_DATABASE=animaldb
```

Alternatively we can use the `--db` command-line parameter

## Backup

To backup a database to a text file, use the `couchbackup` command, directing the output to a text file:

```sh
couchbackup > backup.txt
```

Another way of backing up is to set the `COUCH_URL` environment variable only and supply the database name on the command-line:

```sh
couchbackup --db animaldb > animaldb.txt
```

## Logging & resuming backups

You may also create a log file which records the progress of the backup with the `--log` parameter e.g.

```sh
couchbackup --db animaldb --log animaldb.log > animaldb.txt
```

This log file can be used to resume backups from where you left off with `--resume true`:

```sh
couchbackup --db animaldb --log animaldb.log --resume true >> animaldb.txt
```

The `--resume true` option works for a backup that has finished spooling changes, but has not yet completed downloading all the necessary batches of documents. It does _not_ provide an incremental backup solution.

You may also specify the name of the output file, rather than directing the backup data to *stdout*:

```sh
couchbackup --db animaldb --log animaldb.log --resume true --output animaldb.txt
```

## Restore

Now that we have our backup text file, we can restore it to a new, empty, existing database using the `couchrestore`:

```sh
cat animaldb.txt | couchrestore
```

or specifying the database name on the command-line:

```sh
cat animaldb.txt | couchrestore --db animaldb2
```

## Compressed backups

If we want to compress the backup data before storing to disk, we can pipe the contents through `gzip`:

```sh
couchbackup --db animaldb | gzip > animaldb.txt.gz
```

and restore the file with:

```sh
cat animaldb.tar.gz | gunzip | couchdbrestore --db animaldb2
```

## Encrypted backups

Similarly to compression it is possible to pipe the backup content through an
encryption or decryption utility. For example with `openssl`:

```sh
couchbackup --db animaldb | openssl aes-128-cbc -pass pass:12345 > encrypted_animal.db
```

```sh
openssl aes-128-cbc -d -in encrypted_animal.db -pass pass:12345 | couchrestore --db animaldb2
```

Note that the content is unencrypted while it is being processed by the
backup tool before it is piped to the encryption utility.

## What's in a backup file?

A backup file is a text file where each line contains a JSON encoded array of up to `buffer-size` objects e.g.

```js
    [{"a":1},{"a":2}...]
    [{"a":501},{"a":502}...]
```

## What's in a log file?

A log file contains a line:

- for every batch of document ids that need to be fetched e.g. `:t batch56 [{"id":"a"},{"id":"b"}]`
- for every batch that has been fetched and stored e.g. `:d batch56`
- to indicate that the changes feed was fully consumed e.g. `:changes_complete`

## What is shallow mode?

When you run `couchbackup` with `--mode shallow` a simpler backup is performed, only backing up the winning revisions
of the database. No revision tokens are saved and any conflicting revisions are ignored. This is a faster, but less
complete backup. Shallow backups cannot be resumed because they do not produce a log file.

NOTE: Parallellism will not be in effect if `--mode shallow` is defined.

## Why use CouchBackup?

The easiest way to backup a CouchDB database is to copy the ".couch" file. This is fine on a single-node instance, but when running multi-node
Cloudant or using CouchDB 2.0 or greater, the ".couch" file only contains a single shard of data. This utility allows simple backups of CouchDB
or Cloudant database using the HTTP API.

This tool can be used to script the backup of your databases. Move the backup and log files to cheap Object Storage so that you have multiple copies of your precious data.

## Options reference

### Environment variables

* `COUCH_URL` - the URL of the CouchDB/Cloudant server e.g. `http://127.0.0.1:5984`
* `COUCH_DATABASE` - the name of the database to act upon e.g. `mydb` (default `test`)
* `COUCH_PARALLELISM` - the number of HTTP requests to perform in parallel when restoring a backup e.g. `10` (Default `5`)
* `COUCH_BUFFER_SIZE` - the number of documents fetched and restored at once e.g. `100` (default `500`). When using CouchBackup with [Cloudant on Transaction Engine](https://www.ibm.com/cloud/blog/announcements/ibm-cloudant-on-transaction-engine) `COUCH_BUFFER_SIZE` must be less than `2000` to avoid bad request errors.
* `COUCH_REQUEST_TIMEOUT` - the number of milliseconds to wait for a respose to a HTTP request before retrying the request e.g. `10000` (Default `120000`)
* `COUCH_LOG` - the file to store logging information during backup
* `COUCH_RESUME` - if `true`, resumes a previous backup from its last known position
* `COUCH_OUTPUT` - the file name to store the backup data (defaults to stdout)
* `COUCH_MODE` - if `shallow`, only a superficial backup is done, ignoring conflicts and revision tokens. Defaults to `full` - a full backup.
* `COUCH_QUIET` - if `true`, suppresses the individual batch messages to the console during CLI backup and restore
* `CLOUDANT_IAM_API_KEY` - optional [IAM API key](https://console.bluemix.net/docs/services/Cloudant/guides/iam.html#ibm-cloud-identity-and-access-management)
 to use to access the Cloudant database instead of user information credentials in the URL. The endpoint used to retrieve the token defaults to
 `https://iam.cloud.ibm.com/identity/token`, but can be overridden if necessary using the `CLOUDANT_IAM_TOKEN_URL` environment variable.
* `DEBUG` - if set to `couchbackup`, all debug messages will be sent to `stderr` during a backup or restore process

_Note:_ These environment variables can only be used with the CLI. When
[using programmatically](#using-programmatically) the `opts` dictionary must be
used.

### Command-line parameters

* `--url` - same as `COUCH_URL` environment variable
* `--db` - same as `COUCH_DATABASE`
* `--parallelism` - same as `COUCH_PARALLELISM`
* `--buffer-size` - same as `COUCH_BUFFER_SIZE`
* `--request-timeout` - same as `COUCH_REQUEST_TIMEOUT`
* `--log` - same as `COUCH_LOG`
* `--resume` - same as `COUCH_RESUME`
* `--output` - same as `COUCH_OUTPUT`
* `--mode` - same as `COUCH_MODE`
* `--iam-api-key` - same as `CLOUDANT_IAM_API_KEY`
* `--quiet` - same as `COUCH_QUIET`

## Using programmatically

You can use `couchbackup` programatically. First install
`couchbackup` into your project with `npm install --save @cloudant/couchbackup`.
Then you can import the library into your code:

```js
  const couchbackup = require('@cloudant/couchbackup');
```

The library exports two main functions:

1. `backup` - backup from a database to a writable stream.
2. `restore` - restore from a readable stream to an empty database.

### Examples

See [the examples folder](./examples) for example scripts showing how to
use the library.

### Backup

The `backup` function takes a source database URL, a stream to write to,
backup options and a callback for completion.

```javascript
backup: function(srcUrl, targetStream, opts, callback) { /* ... */ }
```

The `opts` dictionary can contain values which map to a subset of the
environment variables defined above. Those related to the source and
target locations are not required.

* `parallelism`: see `COUCH_PARALLELISM`.
* `bufferSize`: see `COUCH_BUFFER_SIZE`.
* `requestTimeout`: see `COUCH_REQUEST_TIMEOUT`.
* `log`: see `COUCH_LOG`.
* `resume`: see `COUCH_RESUME`.
* `mode`: see `COUCH_MODE`.
* `iamApiKey`: see `CLOUDANT_IAM_API_KEY`.
* `iamTokenUrl`: may be used with `iamApiKey` to override the default URL for
 retrieving IAM tokens.

The callback has the standard `err, data` parameters and is called when
the backup completes or fails.

The `backup` function returns an event emitter. You can subscribe to:

* `changes` - when a batch of changes has been written to log stream.
* `written` - when a batch of documents has been written to backup stream.
* `finished` - emitted once when all documents are backed up.

Backup data to a stream:

```javascript
couchbackup.backup(
  'https://examples.cloudant.com/animaldb',
  process.stdout,
  {parallelism: 2},
  function(err, data) {
    if (err) {
      console.error("Failed! " + err);
    } else {
      console.error("Success! " + data);
    }
  });
```

Or to a file:

```javascript
couchbackup.backup(
  'https://examples.cloudant.com/animaldb',
  fs.createWriteStream(filename),
  {parallelism: 2},
  function(err, data) {
    if (err) {
      console.error("Failed! " + err);
    } else {
      console.error("Success! " + data);
    }
  });
```

### Restore

The `restore` function takes a readable stream containing the data emitted
by the `backup` function and uploads that to a Cloudant database.

_Note:_ A target database must be a **new and empty** database.

```javascript
restore: function(srcStream, targetUrl, opts, callback) { /* ... */ }
```

The `opts` dictionary can contain values which map to a subset of the
environment variables defined above. Those related to the source and
target locations are not required.

* `parallelism`: see `COUCH_PARALLELISM`.
* `bufferSize`: see `COUCH_BUFFER_SIZE`.
* `requestTimeout`: see `COUCH_REQUEST_TIMEOUT`.
* `iamApiKey`: see `CLOUDANT_IAM_API_KEY`.
* `iamTokenUrl`: may be used with `iamApiKey` to override the default URL for
 retrieving IAM tokens.

The callback has the standard `err, data` parameters and is called when
the restore completes or fails.

The `restore` function returns an event emitter. You can subscribe to:

* `restored` - when a batch of documents is restored.
* `finished` - emitted once when all documents are restored.

The backup file (or `srcStream`) contains lists comprising of document
revisions, where each list is separated by a newline. The list length is
dictated by the `bufferSize` parameter used during the backup.

It's possible a list could be corrupt due to failures in the backup process. A
`BackupFileJsonError` is emitted for each corrupt list found. _These can only be
ignored if the backup that generated the stream did complete successfully_. This
ensures that corrupt lists also have a valid counterpart within the stream.

Restore data from a stream:

```javascript
couchbackup.restore(
  process.stdin,
  'https://examples.cloudant.com/new-animaldb',
  {parallelism: 2},
  function(err, data) {
    if (err) {
      console.error("Failed! " + err);
    } else {
      console.error("Success! " + data);
    }
  });
```

Or from a file:

```javascript
couchbackup.restore(
  fs.createReadStream(filename),
  'https://examples.cloudant.com/new-animaldb',
  {parallelism: 2},
  function(err, data) {
    if (err) {
      console.error("Failed! " + err);
    } else {
      console.error("Success! " + data);
    }
  });
```

## Error Handling

The `couchbackup` and `couchrestore` processes are designed to be relatively robust over an unreliable network. Work is batched and any failed requests are retried indefinitely. However, certain aspects of the execution will not tolerate failure:
- Spooling changes from the database changes feed. A failure in the changes request during the backup process will result in process termination.
- Validating the existence of a target database during the database restore process.

### API

When using the library programmatically an `Error` will be passed in one of two ways:
* For fatal errors the callback will be called with `null, error` arguments
* For non-fatal errors an `error` event will be emitted

### CLI Exit Codes

On fatal errors, `couchbackup` and `couchrestore` will exit with non-zero exit codes. This section
details them.

### common to both `couchbackup` and `couchrestore`

* `1`: unknown CLI option or generic error.
* `2`: invalid CLI option.
* `10`: backup source or restore target database does not exist.
* `11`: unauthorized credentials for the database.
* `12`: incorrect permissions for the database.
* `40`: database returned a fatal HTTP error.

### `couchbackup`

* `20`: resume was specified without a log file.
* `21`: the resume log file does not exist.
* `22`: incomplete changes in log file.
* `30`: error spooling changes from the database.
* `50`: source database does not support `/_bulk_get` endpoint.

### `couchrestore`

* `13`: restore target database is not new and empty.

## Note on attachments

TLDR; If you backup a database that contains attachments you will not be able to restore it.

As documented above couchbackup does not support backing up or restoring databases containing documents with attachments.
Attempting to backup a database that includes documents with attachments will appear to succeed. However, the attachment
content will not have been downloaded and the backup file will contain attachment metadata. Consequently any attempt to
restore the backup will result in errors because the attachment metadata will reference attachments that are not present
in the restored database.

It is recommended to store attachments directly in an object store with a link in the JSON document instead of using the
native attachment API.