Managing Workspaces

A Steampipe workspace is a "profile" that allows you to define a unified environment that the Steampipe client can interact with. Each workspace is composed of:

  • a single Steampipe database instance
  • context-specific settings and options (snapshot location, query timeout, etc)

Steampipe workspaces allow you to define multiple named configurations:

workspace "local" {
workspace_database = "local"
}
workspace "acme_prod" {
workspace_database = "acme/prod"
snapshot_location = "acme/prod"
query_timeout = 600
}

and easily switch between them using the --workspace argument or STEAMPIPE_WORKSPACE environment variable:

steampipe query --workspace local "select * from aws_account"
steampipe query --workspace acme_prod "select * from aws_account"

Turbot Pipes workspaces are automatically supported:

steampipe query --workspace acme/dev "select * from aws_account"

Defining Workspaces

Workspace configurations can be defined in any .spc file in the ~/.steampipe/config directory, but by convention they are defined in ~/.steampipe/config/workspaces.spc file. This file may contain multiple workspace definitions that can then be referenced by name.

Any unset arguments will assume the default values - you don't need to set them all:

workspace "default" {
query_timeout = 300
}

You can use base= to inherit settings form another profile:

workspace "dev" {
base = workspace.default
workspace_database = "acme/dev"
}

The workspace_database may be local (which is the default):

workspace "local_db" {
workspace_database = "local"
}

or a Turbot Pipes workspace, in the form of {identity_handle}/{workspace_handle}:

workspace "acme_prod" {
workspace_database = "acme/prod"
}

The snapshot_location can also be a Turbot Pipes workspace, in the form of {identity_handle}/{workspace_handle}:

workspace "acme_prod" {
workspace_database = "acme/prod"
snapshot_location = "acme/prod"
}

If it doesn't match the {identity_handle}/{workspace_handle} pattern it will be interpreted to be a path to a directory in the local filesystem where snapshots should be written to:

workspace "local" {
workspace_database = "local"
snapshot_location = "home/raj/my-snapshots"
}

You can specify options blocks to set options for steampipe query:

workspace "local_dev" {
search_path_prefix = "aws_all"
query_timeout = 300
pipes_token = "tpt_999faketoken99999999_111faketoken1111111111111"
pipes_host = "pipes.turbot.com"
snapshot_location = "acme/dev"
workspace_database = "local"
options "query" {
multi = false # true, false
output = "table" # json, csv, table, line
header = true # true, false
separator = "," # any single char
timing = true # true, false
autocomplete = true
}
}

You can even set the install_dir for a workspace if you want to use the data layer from another Steampipe installation directory.

This allows you to define workspaces that use a database from another installation directory:

workspace "steampipe_2" {
install_dir = "~/steampipe2"
}

and easily switch between them with the --workspace flag:

steampipe query --workspace steampipe_2 "select * from aws_account"

Using Workspaces

Workspaces may be defined in any .spc file in the ~/.steampipe/config directory, but by convention they should be placed in the ~/.steampipe/config/workspaces.spc file.

The workspace named default is special; if a workspace named default exists, --workspace is not specified in the command, and STEAMPIPE_WORKSPACE is not set, then Steampipe uses the default workspace:

steampipe query --snapshot "select * from aws_account"

You can pass any workspace to --workspace to use its values:

steampipe query --snapshot --workspace=acme_dev "select * from aws_account"

Or do the same with the STEAMPIPE_WORKSPACE environment variable:

STEAMPIPE_WORKSPACE=acme_dev steampipe query --snapshot "select * from aws_account"

If you specify the --workspace argument and the STEAMPIPE_WORKSPACE environment variable, the --workspace argument wins:

# acme_prod will be used as the effective workspace
export STEAMPIPE_WORKSPACE=acme_dev
steampipe query --snapshot --workspace=acme_prod "select * from aws_account"

If you specify the --workspace argument and more specific arguments (workspace_database, etc), any more specific arguments will override the workspace values:

# will use "local" as the db, and acme_prod workspace for any OTHER options
steampipe query --snapshot \
--workspace=acme_prod \
--workspace_database=local \
"select * from aws_account"

Environment variable values override default workspace settings when the default workspace is implicitly used:

# will use acme/dev as DB, but get the rest of the values from default workspace
export STEAMPIPE_WORKSPACE_DATABASE=acme/dev
steampipe query "select * from aws_account"

If the default workspace is explicitly passed to the --workspace argument, its values will override any individual environment variables:

# will NOT use acme/dev as DB - will use ALL of the values from default workspace
export STEAMPIPE_WORKSPACE_DATABASE=acme/dev
steampipe query --snapshot --workspace=default "select * from aws_account"

The same is true of any named workspace:

# will NOT use acme/dev as DB - will use ALL of the values from acme_prod workspace
export STEAMPIPE_WORKSPACE_DATABASE=acme/dev
steampipe query --workspace=acme_prod "select * from aws_account"

Implicit Workspaces

Named workspaces follow normal standards for HCL identifiers, thus they cannot contain the slash (/) character. If you pass a value to --workspace or STEAMPIPE_WORKSPACE in the form of {identity_handle}/{workspace_handle}, it will be interpreted as an implicit workspace. Implicit workspaces, as the name suggests, do not need to be specified in the workspaces.spc file. Instead they will be assumed to refer to a Turbot Pipes workspace, which will be used as both the database (workspace_database) and snapshot location (snapshot_location).

Essentially, --workspace acme/dev is equivalent to:

workspace "acme/dev" {
workspace_database = "acme/dev"
snapshot_location = "acme/dev"
}