This pages details how to get Salmon running, either on EC2 or locally on your machine. After you get Salmon running, detail on how to launch experiments in Getting started.


See the Troubleshooting section if you’re having difficulties with either of the processes below.


  1. Sign into Amazon AWS (

  2. Select the “Oregon” region (or us-west-2) in the upper right.

  3. Go to Amazon EC2.

  4. Launch a new instance (the big blue button or square orange button).

  5. Select AMI ami-07d9e17b8dea4da43 titled “Salmon”.

    • Note: the AMI appears in Community AMIs after searching “Salmon” (only in the Oregon/us-west-2 region!).

  6. Select an appropriate instance type.

    • Salmon requires at least 2GB of memory and 1 CPU.

    • t3.large is recommended for passive algorithms (i.e, random sampling).

    • t3.xlarge is recommended for adaptive algorithms (e.g., ARR; see the benchmarks on adaptive algorithm for more detail).

    • Note: is a great resource to check costs. As of April 2022, t3.large and t3.xlarge cost about $2/day and $4/day respectively.

  7. Create a key pair.


    Don’t lose your key pair! Without the key pair, the Salmon developers will be severely limited in the help they can provide.

  8. Don’t click the big orange button yet. Continue to the rules page, and add these rules:

    1. In the networking interface box, check the HTTP and HTTP boxes and select the “edit” button:

    2. After hitting “edit”, scroll down to “add security group rule” and open port 8421 to (aka anyone).

  9. Now, click the big orange button! The AMI will probably take around 15 to initialize (but may take up to 30 minutes).

  10. Keep your “key pair” in a safe place. The key pair typically has a .pem extension.

The AMI initialization is done (which takes about 15 minutes), Salmon will be available at http://[url]:8421. For example, [url] might be the Amazon public DNS or public IP.


By default, Salmon does not support HTTPS. Be sure the URL begins with http:// and not https://!

Until you upload data, http://[url]:8421 will only show an error message. To start using Salmon, these endpoints will be available:

  • http://[url]:8421/init to create a user and initialize a new experiment.

  • http://[url]:8421/docs to see the endpoint documentation. The Salmon version displayed should match the most recent Salmon release in the list of Salmon releases.

  • http://[url]:8421/dashboard to view all relevant links, including links to the…

    • The query page. This is the URL that shows the relevant triplets. This is the URL to be sent to a crowdsourcing service.

    • API documentation. This includes information on how to launch an experiment, and what files need to be uploaded. View the documentation for the POST request /init_exp for more detail.

    • Download the experiment. The downloaded file can be re-uploaded to a new machine so experiments can be restarted.

    • Responses. To get all human responses.

    • Logs. This is very useful for debugging.


    Download all files when stopping or terminating the machine – especially the responses and experiment file.


If you have an issue with the machine running Salmon, be sure to include the logs when contacting the Salmon developers. They’d also appreciate it if you left the machine running.


The storage required for Salmon is 128GB. According to Amazon’s EBS pricing, that costs about $10/month (in April 2022).

Local install

See Install Salmon for the process of installing locally.


See FAQ for more general questions.


Please include the version in any bug reports or feature requests. The version number should look something like v0.4.1. It can be found at http://[url]:8421/docs or in the downloaded experiment file (found at http://[url]:8421/download which has a filename like exp-2021-05-20T07:31-salmon-v0.4.1.rdb).

I can’t access Salmon’s URL

Try using http:// instead of https://. By default, EC2 does not support HTTPS, and some browsers use HTTPS automatically.

I can’t find Salmon’s AMI

Are you in EC2’s Oregon region, us-west-2? That can be changed in the upper right of the Amazon EC2 interface.

The Salmon AMI has been created in the us-west-2 region, and EC2 AMIs are only available in the regions they’re created in.

Restoring from a backup didn’t work

This might happen if Salmon changed between when you downloaded and tried to restore the experiment. Launching from EC2 always downloads the latest version of Salmon, which may not work with your backup file.


Salmon follows semantic software versioning. If the version string in the .rdb file takes the form vA.B.C, then:

Uploading backup files when relevant “backwards incompatible” software changes are made, which should be encoded in the release notes.

So, it uploading your backup did not work (it should if the version numbers are correct), let’s launch the correct version of Salmon’s server on your machine. That requires this process:

  1. Get the correct version of Salmon.

  2. Spin up a Salmon server.

  3. Go to http://localhost:8421/init

  4. Upload the .rdb file to restore.

This process is basically “launch Salmon’s server on your machine.” First, let’s get the right version of Salmon:

$ # Get right version of Salmon
$ git clone
$ cd salmon
$ git checkout v0.7.0  # from .rdb filename; will take the form "vA.B.C" or "vA.B.CrcD"

Second, let’s launch Salmon:

$ docker-compose up  # takes a while
$ # visit http://localhost:8421/init

Now re-upload the file using the interface at the bottom of the screen.

Now Salmon will issue instructions to restart. Let’s do that:

$ # Now, let's follow the directions Salmon gave:
$ docker-compose stop; docker-compose start
$ docker-compose logs -f
$ # visit http://localhost:8421/dashboard