DevOps Guide

Overview

For a general overview of common devops tasks for Synapse services see Synapse Devops Guide - Overview.

Common DevOps Tasks

Specify a Bucket

The gcs:bucket option can be used to set the bucket name, which must be a globally unique name. If the bucket does not exist when Synapse GCS Axon starts up, the gcs:project option must be set to specify the project to create the bucket under.

For example, to set using an environment variable in the docker-compose.yaml file:

environment:
    # ...
    - SYN_GCSAXON_GCS_BUCKET=uniquebucket-123456
    - SYN_GCSAXON_GCS_PROJECT=myproject-123456

Provide GCS Credentials

By default the GCS Axon will use the service-accounts information in Google Compute Engine metadata for generating authorization tokens. For more information about GCE metadata, see the Google Cloud docs.

If you are not running the GCS Axon in GCE, you may provide credentials as a JSON string containing service account or authorized user credentials in the gcs:credentials configuration value, or a path to a file containing service account or authorized user credentials with the gcs:credentials:path configuration value.

An example docker-compose.yaml file can be seen below. This example omits the gcs:credentials configuration. This would cause GCS to resolve the service credentials from GCE metadata:

version: "3.3"
services:
  00.axon:
    user: "999"
    image: vertexproject/synapse-gcs:v1.x.x
    network_mode: host
    restart: unless-stopped
    volumes:
        - ./storage:/vertex/storage
    environment:
        - SYN_GCSAXON_BUCKET: uniquebucket-123456
        - SYN_GCSAXON_PROJECT: myproject-123456
        - SYN_GCSAXON_HTTPS_PORT: null
        - SYN_GCSAXON_AHA_PROVISION: "ssl://aha.<yournetwork>:27272/<guid>?certhash=<sha256>"

Deploy a Mirror

A Synapse GCS Axon mirror can independently read and write to the GCS storage bucket, with all changes to the local index data written back to the upstream.

Inside the AHA container

Generate a one-time use URL for provisioning from inside the AHA container:

python -m synapse.tools.aha.provision.service 01.axon --mirror axon

You should see output that looks similar to this:

one-time use URL: ssl://aha.<yournetwork>:27272/<guid>?certhash=<sha256>

On the Host

Create the container storage directory:

mkdir -p /srv/syn/01.axon/storage
chown -R 999 /srv/syn/01.axon/storage

Create the /srv/syn/01.axon/docker-compose.yaml file with contents:

version: "3.3"
services:
  01.axon:
    user: "999"
    image: vertexproject/synapse-gcs:v1.x.x
    network_mode: host
    restart: unless-stopped
    volumes:
        - ./storage:/vertex/storage
    environment:
        # disable HTTPS API for now to prevent port collisions
        - SYN_GCSAXON_BUCKET: uniquebucket-123456
        - SYN_GCSAXON_PROJECT: myproject-123456
        - SYN_GCSAXON_HTTPS_PORT=null
        - SYN_GCSAXON_AHA_PROVISION=ssl://aha.<yournetwork>:27272/<guid>?certhash=<sha256>

Note

Don’t forget to replace your one-time use provisioning URL!

Start the container:

docker-compose --file /srv/syn/01.axon/docker-compose.yaml pull
docker-compose --file /srv/syn/01.axon/docker-compose.yaml up -d

Devops Details

Docker Images

The Synapse-GCS service is available as a Docker container from Docker Hub. The repository can be found at:

https://hub.docker.com/repository/docker/vertexproject/synapse-gcs/

Note

There are tagged images available on Docker Hub which correspond to software releases seen in the changelog. The docker tag master is the latest development release. A generic major version tag is available, representing the latest release on a given major version. For example, the v2.x.x tag represents the most current release for the v2.x.x release line. You can utilize specific tagged versions, or a major version specifier, depending on your chosen deployment strategy.

Required Permissions

The provided credentials must have the following permissions:

storage.buckets.get: This is used to ensure that the GCS bucket used to store files in exists.
storage.objects.get: This is used to retrieve files uploaded to the GCS bucket.
storage.objects.list: This is used to list the files in the GCS bucket for index recovery operations.
storage.objects.create: This is used to put files in the GCS bucket.
storage.objects.delete: This is used to delete files uploaded to the GCS bucket. In the case where a bulk delete is requested the delete operation is run on each object. If a GCS delete operation fails (e.g. a role permission error), the SHA-256 will still be removed from the Axon index and the object would need to be manually deleted in the bucket.
storage.buckets.create: This permission is only needed to make the bucket if the bucket does not exist. It is best practice when integrating with GCS to create the bucket separately, since the Synapse GCS service does not specify any sort of access policies when it creates the bucket.

Since Synapse GCS Axon implements the Synapse Axon API, permissions are also inherited. For example, the axon.get rule is required to retrieve a file and the axon.upload rule is required to upload a file.

Configuration Options

The following is a list of available configuration options.

aha:admin

An AHA client certificate CN to register as a local admin user.

Type: string
Environment Variable: SYN_GCSAXON_AHA_ADMIN

aha:leader

The AHA service name to claim as the active instance of a storm service.

Type: string
Environment Variable: SYN_GCSAXON_AHA_LEADER

aha:name

The name of the cell service in the aha service registry.

Type: string
Environment Variable: SYN_GCSAXON_AHA_NAME

aha:network

The AHA service network. This makes aha:name/aha:leader relative names.

Type: string
Environment Variable: SYN_GCSAXON_AHA_NETWORK

aha:provision

The telepath URL of the aha provisioning service.

Type: ['string', 'array']
Environment Variable: SYN_GCSAXON_AHA_PROVISION

aha:registry

The telepath URL of the aha service registry.

Type: ['string', 'array']
Environment Variable: SYN_GCSAXON_AHA_REGISTRY

aha:user

The username of this service when connecting to others.

Type: string
Environment Variable: SYN_GCSAXON_AHA_USER

auth:anon

Allow anonymous telepath access by mapping to the given user name.

Type: string
Environment Variable: SYN_GCSAXON_AUTH_ANON

auth:passwd

Set to <passwd> (local only) to bootstrap the root user password.

Type: string
Environment Variable: SYN_GCSAXON_AUTH_PASSWD

backup:dir

A directory outside the service directory where backups will be saved. Defaults to ./backups in the service storage directory.

Type: string
Environment Variable: SYN_GCSAXON_BACKUP_DIR

dmon:listen

A config-driven way to specify the telepath bind URL.

Type: ['string', 'null']
Environment Variable: SYN_GCSAXON_DMON_LISTEN

gcs:bucket

The name of the bucket to use (must be a globally unique name).

Type: string
Environment Variable: SYN_GCSAXON_GCS_BUCKET

gcs:credentials

Service account or authorized user credentials blob.

Type: object
Environment Variable: SYN_GCSAXON_GCS_CREDENTIALS

gcs:credentials:path

Path to a file containing service account or authorized user credentials.

Type: string
Environment Variable: SYN_GCSAXON_GCS_CREDENTIALS_PATH

gcs:nocredentials

Disable authorization tokens (used for connecting to an emulated GCS server).

Type: boolean
Default Value: False
Environment Variable: SYN_GCSAXON_GCS_NOCREDENTIALS

gcs:project

The name of the project to create the bucket in if it does not exist.

Type: string
Environment Variable: SYN_GCSAXON_GCS_PROJECT

gcs:retry

Maximum number of retries for failed connection attempts.

Type: integer
Default Value: 5
Environment Variable: SYN_GCSAXON_GCS_RETRY

gcs:ssl

Set to false to disable ssl verification when connecting to Google Cloud Storage.

Type: boolean
Default Value: True
Environment Variable: SYN_GCSAXON_GCS_SSL

gcs:url

The base url to use when connecting to Google Cloud Storage.

Type: string
Default Value: 'https://www.googleapis.com'
Environment Variable: SYN_GCSAXON_GCS_URL

http:proxy

An aiohttp-socks compatible proxy URL to use in the wget API.

Type: string
Environment Variable: SYN_GCSAXON_HTTP_PROXY

https:headers

Headers to add to all HTTPS server responses.

Type: object
Environment Variable: SYN_GCSAXON_HTTPS_HEADERS

https:parse:proxy:remoteip

Enable the HTTPS server to parse X-Forwarded-For and X-Real-IP headers to determine requester IP addresses.

Type: boolean
Default Value: False
Environment Variable: SYN_GCSAXON_HTTPS_PARSE_PROXY_REMOTEIP

https:port

A config-driven way to specify the HTTPS port.

Type: ['integer', 'null']
Environment Variable: SYN_GCSAXON_HTTPS_PORT

limit:disk:free

Minimum disk free space percentage before setting the cell read-only.

Type: ['integer', 'null']
Default Value: 5
Environment Variable: SYN_GCSAXON_LIMIT_DISK_FREE

max:bytes

The maximum number of bytes that can be stored in the Axon.

Type: integer
Environment Variable: SYN_GCSAXON_MAX_BYTES

max:count

The maximum number of files that can be stored in the Axon.

Type: integer
Environment Variable: SYN_GCSAXON_MAX_COUNT

max:users

Maximum number of users allowed on system, not including root or locked/archived users (0 is no limit).

Type: integer
Default Value: 0
Environment Variable: SYN_GCSAXON_MAX_USERS

nexslog:en

Record all changes to a stream file on disk. Required for mirroring (on both sides).

Type: boolean
Default Value: False
Environment Variable: SYN_GCSAXON_NEXSLOG_EN

onboot:optimize

Delay startup to optimize LMDB databases during boot to recover free space and increase performance. This may take a while.

Type: boolean
Default Value: False
Environment Variable: SYN_GCSAXON_ONBOOT_OPTIMIZE

tls:ca:dir

An optional directory of CAs which are added to the TLS CA chain for wget and wput APIs.

Type: string
Environment Variable: SYN_GCSAXON_TLS_CA_DIR