Space Daemon
Space Daemon is a wrapper built in Go around awesome IPFS tools so that you can have start coding a decentralized desktop app as fast as possible. It's built on top of Textile Threads and Buckets. Out of the box it includes:
-
A running local instance of Textile Threads.
-
Interfaces to create local private, encrypted buckets.
-
Interfaces for sharing those buckets and the files within.
-
Identity service so that sharing can be done through usernames or emails.
-
FUSE for drive mounting, so that the files can be explored natively in your OS.
-
Key management.
Note: This project is in active development, so it might change its API until it reaches a stable version.
Installation
By default, Space Daemon connects to hosted services provided by Fleek. This should be good if you just want to get it running quickly. However, if you want to connect to your own services, read the Modules Section.
Downloading the binary
Check out the releases here. You can download the latest version for your OS and you should be good to go.
If you want to run Space Daemon by source, check out this section
Usage
Space Daemon provides a gRPC interface. You can read its proto schema here. It contains methods to:
-
Create files and directories
-
List files and directories
-
Creating buckets
-
Sharing buckets
-
Creating identities
You can also use the JavaScript client here https://github.com/FleekHQ/space-client
This can be useful if, for example, you are building a web app that needs to interact with a user's locally running Space Daemon.
Modules
Space Daemon requires a few modules to run successfully. If you downloaded the binary, you don't have to worry about this since it will be connecting to our services. It's good to understand what's happening behind the scenes though.
IPFS Node
All encrypted files are stored in an IPFS node. For convenience, Space Daemon runs an embedded node within the daemon that can be configured as well as the option to specify an external node to connect to.
If you have your own node outside of the daemon, then set the flag -ipfsnode
to false
. This will not spin up an embedded node. You can then connect to your external node by providing the -ipfsaddr
flag (e.g. -ipfsaddr=/ip4/127.0.0.1/tcp/5001
).
In the case you are running the embedded IPFS node, you can further configure the listen address and data directory by setting these flags respectively: -ipfsnodeaddr
and -ipfsnodepath
.
Textile Hub
Required for sharing files between users and backing it up. It stores all backed up files encrypted using a set of keys so that only you, and people you share files with, can read the data. We host our own instance of the Textile Hub, and by default, Space Daemon will conect to it. It can be customized by providing the -textilehub
flag and -textilethreads
flag.
If you want to host your own Textile Hub node, you can read its documentation here
Space Services
We provide hosted alternatives for these services. You can deploy your own by following the instructions in its repo:
https://github.com/fleekHQ/space-services
Identity
These are centralized services that are optional, but offer additional convenience. Used mainly for identity. By using these services, you can allow users to claim usernames, so that Space Daemon can know the public key of a given username and in that way share files via username without having to input public keys directly.
Authentication
Our hosted Textile Hub requires authentication via public key for logging in. This service sends a challenge to Space Daemon, which signs the challenge with the private key of the user and in that way our hosted Textile Hub can allow the user to store data.
Running from source
After cloning this repo, you can run it from source by running go run ./cmd/space-daemon -dev
. Consider that you will need the following environment variables exported in your system:
IPFS_ADDR=[Your IPFS node address]
SERVICES_API_URL=[The URL where Space Services API is located]
VAULT_API_URL=[The URL where Space Vault API is located]
VAULT_SALT_SECRET=[A random string used for kdf functions before storing keys to the vault]
SERVICES_HUB_AUTH_URL=[The URL where Space Services Textile Hub Authorizer is located]
TXL_HUB_TARGET=[The URL of the Textile Hub]
TXL_HUB_MA=[The multiaddress for the Textile hub]
TXL_THREADS_TARGET=[The URL of the Textile Hub where Threads are hosted, can be the same that TXL_HUB_TARGET]
# NOTE: the following are required temporarily and will be removed once hub auth wrapper is setup
TXL_USER_KEY=[Space level key for hub access]
TXL_USER_SECRET=[Space level secret for hub access]
Alternatively, you can run make
to compile the binary. Make sure you have these environment variables exposed though. You can see some example environment variables in .env.example
.
Contributting
We are happy to receive issues and review pull requests. Please make sure to write tests for the code you are introducing and make sure it doesn't break already passing tests.
Read the following sections for an introduction into the code.
Package Structure
Loosely based on these resources: https://github.com/golang-standards/project-layout
/grpc
Folder structure for gRPC and REST API./cmd
Entry point directory for all binaries this repo handles. E.g cmd/{binary-name}/main.go/config
Global Config code/core
Directory for the core objects of the package/logger
Directory for app logging/examples
Directory playground for general examples and drafts
Main classes
ipfs
: contains utils for general IPFS operations.keychain
: manages user public/private key pair.libfuse
: interoperates with FUSE for mounting drives.space
: contains the main integration from the services to the final Textile or FS operations.store
: contains a wrapper around a local db.sync
: keeps track of open files so that the updates get pushed to IPFStextile
: wrapper around Textile booting and operations
Generating Mocks
Mocks are generated using https://github.com/vektra/mockery.
For Linux it needs to be built from source.
mockery --name InterfaceToMock --dir path/to/go/files
Protobuf
If you update the gRPC API, you need to regenerate the Protobuf file.
You will need to install the following binaries in your Go path:
go get -u github.com/grpc-ecosystem/grpc-gateway/protoc-gen-grpc-gateway
Checking the binaries:
ls $GOPATH/bin
Should show the following binaries in your path: protoc-gen-go, protoc-gen-grpc-gateway
Run the protobuf generation:
make proto_gen
Run the REST proxy generation:
make gen_rest
** Ideally you should run make gen_all
before commiting as this would run all the above three code generations and
ensure everything is up to date **
NOTE: See here for instructions on Reverse Proxy: https://github.com/grpc-ecosystem/grpc-gateway
Debugging, Profiling and Tracing
The following flags can be run with the binary to output profiling files for debugging.
Flags support a full path to a file.
-cpuprofile cpu.prof -memprofile mem.prof
By default, the binary runs in debug mode (this may change after release) and it boots a pprof server in localhost:6060. See docs how to interact with pprof server here: https://github.com/google/pprof/blob/master/doc/README.md
To disable debug mode add this flag to binary arguments
-debug=false
To enable trace in the daemon, pass -trace
to the binary arguments. The daemon uses jaegar
for collecting trace information. Run make jaegar
to quickly start a jaeger agent that collects the daemons trace information.
You can http://localhost:16686/
to explore the web ui for traces collected.
CI Secrets
Secrets are set by adding them in Github and then specifying them in release.yml
. Secrets can be constant across environment/stages or be stage specific.
If specified, the release file will dynamically generate the secret name based on the stage by adding a _DEV
or _PRD
suffix to the secret name only for the specificed environment variable. It will always use _PRD
unless the tag ends in -dev
. So for example tag v0.0.15
will use PRD values, while v0.0.15-dev
will use DEV values.
Stage specific secret names will only be used for secrets in release.yml
that point to the step output instead of the secret name directly (i.e., SERVICES_API_URL: ${{ secrets[steps.secretnames.outputs.SERVICES_API_URL] }}
instead of SERVICES_API_URL: ${{ secrets.SERVICES_API_URL }}
.
So to add a new secret:
- If it's not stage specific then add the secret in GH with no suffix and in
release.yml
, refer to it based on the secret name. - If it is stage specific, then create the 2 secrets in GH (ending in
_PRD
and_DEV
), add the entry in stepsecretnames
, and make sure the secret name in the next step points to the step output
评论