Skip to content

Latest commit

 

History

History
366 lines (288 loc) · 15.6 KB

README.md

File metadata and controls

366 lines (288 loc) · 15.6 KB
Raw

MRD Storage Server

The MRD Storage Server provides a simple RESTful API for storing and retrieving data during MRI image reconstructions. It supports:

  • Creating a blob
  • Reading a blob
  • Searching for blobs
  • Retrieving the latest blob matching a search expression

Tags

Blobs are described with a set of tags and searches are expressed as a set of filters over the tags. These tags are specified using URI query string parameters both when creating and searching blobs.

The tags are:

Tag Description
subject The patient ID. Must be specified when creating and searching. For data that is not related to a patient, specify $null.
device The device or scanner ID.
session The session ID.
name The name of the blob.
[custom-tag] Any number of custom tags. Unlike system tags, custom tags can have multiple values.

Tag names are case-insensitive, but their values are case-sensitive.

Blob Metadata

The metadata stored with each blob includes its tags and a number of other properties. This metadata is returned as a JSON document on creates, metadata reads, and searches. When a blob's data is read, the metadata is returned as HTTP response headers. This metadata is:

JSON Field HTTP Header Description
[tag-name] Mrd-Tag-[tag-name] The value of each tag. For custom tags with multiple values, the json field value will be an array of strings and the HTTP header will be a comma-delimited string.
contentType Content-Type The blob's MIME type, using the standard HTTP header for creates and reads.
lastModified Last-Modified The blob's creation timestamp. Using the standard HTTP header, even though blobs are immutable.
expires Expires A datetime after which the blob will be deleted, if the blob was created with a _ttl=duration query parameter.
location Location A URI for reading the blob metadata. System-assigned and globally unique. [base]/v1/blobs/{{id}}
data N/A A URI for reading the blob data. System-assigned and globally unique. [base]/v1/blobs/{{id}}/data

API

Creating a Blob

Tag values are specified as query string parameters of a POST request:

POST http://localhost:3333/v1/blobs/data?subject=123&session=mysession&name=NoiseCovariance
Content-Type: text/plain

This is my content

Response:

HTTP/1.1 201 Created
Date: Fri, 05 Nov 2021 10:51:54 GMT
Content-Length: 310
Content-Type: text/plain; charset=utf-8

{
   "contentType":"text/plain",
   "data":"http://localhost:3333/v1/blobs/c8a3aa43-04c0-4acb-9154-ce7b281ec274-123/data",
   "lastModified":"2021-11-05T11:51:54.036+01:00",
   "location":"http://localhost:3333/v1/blobs/c8a3aa43-04c0-4acb-9154-ce7b281ec274-123",
   "name":"NoiseCovariance",
   "session":"mysession",
   "subject":"123"
}

Blobs can be created with a time to live (TTL). Once the TTL expires, the blob is deleted. The TTL is specified with a _ttl=[duration] query parameter, where the duration is a sequence of decimal numbers, each with optional fraction and a unit suffix, for example 60m or 2h45m. Valid time units are "s" (seconds), "m" (minutes), and "h" (hours).

POST http://localhost:3333/v1/blobs/data?subject=123&session=mysession&name=NoiseCovariance&_ttl=48h
Content-Type: text/plain

This is my content

Response:

HTTP/1.1 201 Created
X-Request-Id: 110c8c60-6ec1-4fa2-8649-a85479a8df43
Date: Fri, 18 Mar 2022 15:43:04 GMT
Content-Length: 342
Content-Type: text/plain; charset=utf-8
Connection: close

{
  "contentType": "text/plain",
  "data": "http://localhost:3333/v1/blobs/a80f86b2-c8c2-4d53-a394-aa548a0a8d34-123/data",
  "expires": "2022-03-20T15:43:04.152Z",
  "lastModified": "2022-03-18T15:43:04.152Z",
  "location": "http://localhost:3333/v1/blobs/a80f86b2-c8c2-4d53-a394-aa548a0a8d34-123",
  "name": "NoiseCovariance",
  "session": "mysession",
  "subject": "123"
}

Note the expires field is 48 hours after the lastModified value.

Reading a Blob

The data attribute in the POST response body contains a link where you can GET the blob content:

GET http://localhost:3333/v1/blobs/c8a3aa43-04c0-4acb-9154-ce7b281ec274-123/data

Response:

HTTP/1.1 200 OK
Content-Type: text/plain
Last-Modified: Fri, 05 Nov 2021 11:51:54 GMT
Mrd-Tag-Name: NoiseCovariance
Mrd-Tag-Session: mysession
Mrd-Tag-Subject: 123
Date: Fri, 05 Nov 2021 10:54:30 GMT
Content-Length: 18

This is my content

Note that the tag values are added as HTTP headers with the prefix Mrd-Tag-.

Searching Blobs:

You can search for blobs based on tags using the same syntax that is used for creating blobs:

GET http://localhost:3333/v1/blobs?subject=123&session=mysession&name=NoiseCovariance

Response:

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Fri, 05 Nov 2021 10:55:26 GMT
Content-Length: 322

{
  "items": [
    {
      "contentType": "text/plain",
      "data": "http://localhost:3333/v1/blobs/c8a3aa43-04c0-4acb-9154-ce7b281ec274-123/data",
      "lastModified": "2021-11-05T11:51:54.036+01:00",
      "location": "http://localhost:3333/v1/blobs/c8a3aa43-04c0-4acb-9154-ce7b281ec274-123",
      "name": "NoiseCovariance",
      "session": "mysession",
      "subject": "123"
    }
  ]
}

Items are sorted in descending order of creation. If not all results fit in a single response, there will be a nextLink field in the response:

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Fri, 05 Nov 2021 10:57:43 GMT
Content-Length: 458

{
  "items": [
    {
      "contentType": "text/plain",
      "data": "http://localhost:3333/v1/blobs/b8b1cac9-f9e6-4f86-b6c6-5eb6bd6cef2a-123/data",
      "lastModified": "2021-11-05T11:57:16.605+01:00",
      "location": "http://localhost:3333/v1/blobs/b8b1cac9-f9e6-4f86-b6c6-5eb6bd6cef2a-123",
      "name": "NoiseCovariance",
      "session": "mysession",
      "subject": "123"
    }
  ],
  "nextLink": "http://localhost:3333/v1/blobs?_ct=eyJ0cyI6MTYzNjEwOTgzNjYwNX0&_limit=1&name=NoiseCovariance&session=mysession&subject=123"
}

It is also possible to only get results that were created at or before a specific time with the _at parameter. The _at parameter is specified as a time zone offset string. For example:

GET http://localhost:3333/v1/blobs?subject=123&session=mysession&name=NoiseCovariance&_at=2021-10-19T15:07:17.224Z

This will exclude results that were created after the given time.

Getting the Latest Blob Matching a Query

There is a shortcut to get the latest blob matching a search query in one request at the /v1/blobs/latest endpoint:

GET http://localhost:3333/v1/blobs/data/latest?subject=123&session=mysession&name=NoiseCovariance

Response:

HTTP/1.1 200 OK
Content-Type: text/plain
Last-Modified: Fri, 05 Nov 2021 11:57:16 GMT
Location: http://localhost:3333/v1/blobs/b8b1cac9-f9e6-4f86-b6c6-5eb6bd6cef2a-123
Mrd-Tag-Name: NoiseCovariance
Mrd-Tag-Session: mysession
Mrd-Tag-Subject: 123
Date: Fri, 05 Nov 2021 11:03:58 GMT
Content-Length: 18

This is my content

The _at parameter can be used here as well to request the latest blob that was created no later than a given time.

GET http://localhost:3333/v1/blobs/data/latest?subject=123&session=mysession&name=NoiseCovariance&_at=2021-10-19T15:07:17.224Z

Custom tags

Custom tags can be provded for blobs. Unlike system tags, custom tags can have many values:

POST http://localhost:3333/v1/blobs/data?subject=someone&customTag1=a&customTag1=b

HTTP/1.1 201 Created
Date: Fri, 05 Nov 2021 11:05:44 GMT
Content-Length: 298
Content-Type: text/plain; charset=utf-8

{"contentType":"text/plain","customtag1":["a","b"],"data":"http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone/data","lastModified":"2021-11-05T12:05:44.037+01:00","location":"http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone","subject":"someone"}

When a custom tag has multiple values, searches match any of the values:

GET http://localhost:3333/v1/blobs?subject=someone&customTag1=a

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Fri, 05 Nov 2021 11:06:58 GMT
Content-Length: 310

{
  "items": [
    {
      "contentType": "text/plain",
      "customtag1": [
        "a",
        "b"
      ],
      "data": "http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone/data",
      "lastModified": "2021-11-05T12:05:44.037+01:00",
      "location": "http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone",
      "subject": "someone"
    }
  ]
}

However, when specifying multiple criteria on a tag, they must all match (they are ANDed togther, not ORed):

GET http://localhost:3333/v1/blobs?subject=someone&customTag1=a&customTag1=missing

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Fri, 05 Nov 2021 11:07:38 GMT
Content-Length: 13

{
  "items": []
}

GET http://localhost:3333/v1/blobs?subject=someone&customTag1=a&customTag1=b

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Fri, 05 Nov 2021 11:08:05 GMT
Content-Length: 310

{
  "items": [
    {
      "contentType": "text/plain",
      "customtag1": [
        "a",
        "b"
      ],
      "data": "http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone/data",
      "lastModified": "2021-11-05T12:05:44.037+01:00",
      "location": "http://localhost:3333/v1/blobs/35d4ae47-6e1c-4881-adb9-5581cededcbb-someone",
      "subject": "someone"
    }
  ]
}

Health Check Endpoint

There is a /healthcheck endpoint that can be used to verify that that the server is functioning correctly.

GET http://localhost:3333/healthcheck

Example response when the server is in a healthy state:

HTTP/1.1 200 OK
Content-Type: application/json; charset=utf-8
Date: Wed, 09 Mar 2022 17:12:13 GMT
Content-Length: 16

{
  "status": "OK"
}

Example when there is an issue:

HTTP/1.1 503 Service Unavailable
Content-Type: application/json; charset=utf-8
Date: Wed, 09 Mar 2022 17:13:03 GMT
Content-Length: 87

{
  "status": "Service Unavailable",
  "errors": {
    "database":"failed to connect to database"
  }
}

Error details are written to the log (stderr).

Data Store Providers

Blob Metadata (tags) are stored separately from the blob contents. We currently support PostgreSQL and SQLite for the metadata and the filesystem or Azure Blob Storage for storing blob contents.

Getting Started

By default, the storage server uses SQLite and the filesystem. The behavior of the server can be configured using environment variables:

Variable Type Description Default Value
MRD_STORAGE_SERVER_DATABASE_PROVIDER string The metadata database provider. Can be sqlite or postgresql. sqlite
MRD_STORAGE_SERVER_DATABASE_CONNECTION_STRING string The provider-specific connection string. For SQLite, the path to the database file. ./data/metadata.db
MRD_STORAGE_SERVER_DATABASE_PASSWORD string If specified, provides a password that will be added to the PostgreSQL connection string. Appends password=<value> to the connection string. The connection string must be given in keyword/value format, not as a URI. ./data/metadata.db
MRD_STORAGE_SERVER_STORAGE_PROVIDER string The blob storage provider. Can be filesystem or azureblob. filesystem
MRD_STORAGE_SERVER_STORAGE_CONNECTION_STRING string The provider-specific connection string. For the filesystem provider, the path to the directory in which to store files. ./data/blobs
MRD_STORAGE_SERVER_STORAGE_PORT integer The port to listen on. 3333
MRD_STORAGE_SERVER_STORAGE_LOG_REQUESTS boolean Whether to log the URI, status code, and duration of each HTTP request. true

In addition, any of the above values can be provided as a file instead of being stored in environment variables, where they could end up exposed by logging tools. To do this, append _FILE to the environment variable name and provide the file path as the value. For example, you can write the database connection string to a file, and set the following environment variable pointing to the file:

export MRD_STORAGE_SERVER_DATABASE_CONNECTION_STRING_FILE="/path/to/the/connection_string_file.txt"