Integrations FAQ¶

What’s the difference between incoming and outgoing webhooks?¶

A webhook is a way for one app to send real-time data to another app.

In Mattermost, incoming webhooks receive data from external applications and make a post in a specified channel. They’re great for setting up notifications when something happens in an external application.

Outgoing webhooks take data from Mattermost, and send it to an external application. Then the outgoing webhook can post a response back in Mattermost. They’re great for listening in on channels, and then notifying external applications when a trigger word is said.

What is a slash command?¶

A slash command is similar to an outgoing webhook, but instead of listening to a channel it is used as a command tool. This means if you type in a slash command it will not be posted to a channel, whereas an outgoing webhook is only triggered by posted messages.

What does Slack-compatible mean?¶

Slack compatible means that Mattermost accepts integrations that have a payload in the same format as Slack.

If you have a Slack integration, you should be able to set it up in Mattermost without changing the format.

What if I have a webhook from somewhere other than Slack?¶

If you have an integration that outputs a payload in a different format you need to write an intermediate application to act as a translation layer to change it to the format Mattermost uses. Since there’s currently no general standard for webhook formatting, this is unavoidable and just a part of how webhooks work.

If there’s no translation layer, Mattermost won’t understand the data you’re sending.

What are attachments?¶

When “attachments” are mentioned in the integrations documentation, it refers to Slack’s Message Attachments. These “attachments” can be optionally added as an array in the data sent by an integration, and are used to customize the formatting of the message.

We currently don’t support the ability to attach files to a post made by an integration.

Where can I find existing integrations?¶

Visit our app directory for dozens of open source integrations to common tools like Jira, Jenkins, GitLab, Trac, Redmine, and Bitbucket, along with interactive bot applications (Hubot, mattermost-bot), and other communication tools (Email, IRC, XMPP, Threema) that are freely available for use and customization.

Where should I install my integrations?¶

For self-hosted deployments in small setups you might host integrations on the same server on which Mattermost is installed. For larger deployments you can setup a separate server for integrations, or add them to the server on which the external application is hosted–for example, if you’re self-hosting a Jira server you could deploy a Jira integration on the Jira server itself.

When self-hosting restrictions are less strict, AWS, Heroku and other public cloud options could also be used.

How do I create a bot account without using webhooks?¶

Deployments that cannot create bot accounts via webhooks due to security reasons can use the following approach:

Create a bot account using a secure email and strong password.
Manually add the account to all teams and channels it needs access to. If your deployment has a lot of teams or channels, you may create a CLI script to automate the process.
- In a testing environment, you may also make the bot account a System Admin, giving the bot permissions to post to any channel. Not recommended in production due to potential security vulnerabilities.
Provide the email and password to your integration, and store it in a secure location with restricted access.
Have your integration use the email and password with an /api/v4/login endpoint to retrieve a session token. The session token is used to authenticate to the Mattermost system.
- Set up your bot to make an HTTP POST to your-mattermost-url.com/api/v4/users/login with a JSON body, including the bot account’s email and password.
```
POST /api/v4/users/login HTTP/1.1
Host: your-mattermost-url.com
Content-Length: 66
Content-Type: application/json

{"login_id":"someone@nowhere.com","password":"thisisabadpassword"}
```
  where we assume there is a Mattermost instance running at http://localhost:8065.
- If successful, the response will contain a Token header and a user object in the body:
```
HTTP/1.1 200 OK
Set-Cookie: MMSID=hyr5dmb1mbb49c44qmx4whniso; Path=/; Max-Age=2592000; HttpOnly
Token: hyr5dmb1mbb49c44qmx4whniso
X-Ratelimit-Limit: 10
X-Ratelimit-Remaining: 9
X-Ratelimit-Reset: 1
X-Request-Id: smda55ckcfy89b6tia58shk5fh
X-Version-Id: developer
Date: Fri, 11 Sep 2015 13:21:14 GMT
Content-Length: 657
Content-Type: application/json; charset=utf-8

{{user object as json}}
```
  The bot should retrieve the session token from the Token header and store it in memory for use with future requests.
- Note: Each session token has an expiry time, set depending on the server’s configuration. If the session token your bot is using expires, it will receive a 401 Unauthorized response from requests using that token. When your bot receives this response, it should re-apply the login logic (using the above steps) to get another session token. Then re-send the request that received the 401 status code.
Include the Token as part of the Authorization header on API requests from your integration.
- To confirm the token works, you can have your bot make a simple GET request to /api/v4/users/me with the Authorization: bearer <yourtokenhere> in the header. If it returns a 200 with the bot’s user object in the response, the API request was made successfully.
```
GET /api/v4/users/me HTTP/1.1
Authorization: bearer <yourtokenhere>
Host: your-mattermost-url.com
```

Note: The Mattermost development team is also working on an API developer token, which allows you to authenticate the bot account via the API token rather than retrieving a session token from a user account.

How should I automate the install and upgrade of Mattermost when included in another application?¶

Automating Mattermost installation within another application:

Review the Mattermost installation guides to understand configuration steps of the production deployment
Install Mattermost files to a dedicated /opt/mattermost directory by decompressing the tar.gz file of the latest release for your target platform (for example linux-amd64).
Review Configuration Settings in config.json and set your automation to customize your Mattermost deployment based on your requirements.
For directory locations defined in config.json, such as the location of the local file storage directory (./data/) or logs directory (./logs), you can redefine those locations in your config.json settings and move the directories.
- All other directories should remain as they are in /mattermost
Test that your Mattermost server is running with your new configuration.
Also, from the commandline run ./bin/platform -version to test that the commandline interface is functioning properly.

Automating Mattermost upgrade within another application:

Review the upgrade guide for an overview of the upgrade procedure.
Create automation to upgrade to the next Mattermost versions:
- backup the config.json file to preserve any settings a user may have made.
- backup the ./data directory if local storage is used for files.
- replace the contents of /mattermost directory with the decompressed contents of the latest release.
- restore config.json and ./data to their previous locations (which may have been overwritten).
- if you need to overwrite any config.json parameters use a sed command or similar tool to update config.json
- starting the Mattermost server to upgrade the database, config.json file, and ./data as necessary.
Optionally the upgrade procedure can be chained so users can upgrade across an arbitrary number of Mattermost versions rather than to just the latest release.