English | 简体ä¸ć–‡
A cross-platform real-time file synchronization tool out of the box based on Golang.
The first need Go installed (version 1.22+ is required), then you can use the below
command to install gofs
.
go install github.com/no-src/gofs/...@latest
You can use the build-docker.sh script to build the docker image and you should clone this
repository and cd
to the root path of the repository first.
$ ./scripts/build-docker.sh
Or pull the docker image directly from DockerHub with the command below.
$ docker pull nosrc/gofs
For more scripts about release and docker, see the scripts directory.
You can install a program run in the background using the following command on Windows.
go install -ldflags="-H windowsgui" github.com/no-src/gofs/...@latest
Please ensure the source directory and dest directory exists first, replace the following path with your real path.
$ mkdir source dest
Generate the TLS cert file and key file for testing purposes.
The TLS cert and key files are just used by File Server and Remote Disk Server.
$ go run $GOROOT/src/crypto/tls/generate_cert.go --host 127.0.0.1
2021/12/30 17:21:54 wrote cert.pem
2021/12/30 17:21:54 wrote key.pem
Look up our workspace.
$ ls
cert.pem key.pem source dest
Synchronize files between disks by Local Disk.
sequenceDiagram
participant DA as DiskA
participant C as Client
participant DB as DiskB
autonumber
C ->> DA: monitor disk
DA ->> C: notify change
C ->> DA: read file
DA ->> C: return file
C ->> DB: write file
Synchronize files from server by Remote Disk Server and Remote Disk Client.
sequenceDiagram
participant SD as Server Disk
participant S as Server
participant C as Client
participant CD as Client Disk
autonumber
S ->> SD: monitor disk
C ->> S: connect and auth
SD ->> S: notify change
S ->> C: notify change
C ->> S: pull file
S ->> SD: read file
SD ->> S: return file
S ->> C: send file
C ->> CD: write file
Synchronize files to server by Remote Push Server and Remote Push Client.
sequenceDiagram
participant CD as Client Disk
participant C as Client
participant S as Server
participant SD as Server Disk
autonumber
C ->> CD: monitor disk
CD ->> C: notify change
C ->> CD: read file
CD ->> C: return file
C ->> S: push file
S ->> SD: write file
Synchronize files from SFTP server by SFTP Pull Client.
sequenceDiagram
participant CD as Client Disk
participant C as Client
participant SS as SFTP Server
participant SSD as SFTP Server Disk
autonumber
C ->> SS: pull file
SS ->> SSD: read file
SSD ->> SS: return file
SS ->> C: send file
C ->> CD: write file
Synchronize files to SFTP server by SFTP Push Client.
sequenceDiagram
participant CD as Client Disk
participant C as Client
participant SS as SFTP Server
participant SSD as SFTP Server Disk
autonumber
C ->> CD: monitor disk
CD ->> C: notify change
C ->> CD: read file
CD ->> C: return file
C ->> SS: push file
SS ->> SSD: write file
Synchronize files from MinIO server by MinIO Pull Client.
sequenceDiagram
participant CD as Client Disk
participant C as Client
participant MS as MinIO Server
participant MSD as MinIO Server Disk
autonumber
C ->> MS: pull file
MS ->> MSD: read file
MSD ->> MS: return file
MS ->> C: send file
C ->> CD: write file
Synchronize files to MinIO server by MinIO Push Client.
sequenceDiagram
participant CD as Client Disk
participant C as Client
participant MS as MinIO Server
participant MSD as MinIO Server Disk
autonumber
C ->> CD: monitor disk
CD ->> C: notify change
C ->> CD: read file
CD ->> C: return file
C ->> MS: push file
MS ->> MSD: write file
Start a Task Client to subscribe to the Task Server, then acquire the task and execute it, take the From Server for example.
sequenceDiagram
participant A as Admin
participant TS as Task Server
participant SD as Server Disk
participant S as Server
participant TC as Task Client
participant CW as Client Worker
participant CD as Client Disk
participant TQ as Task Queue
autonumber
S ->> SD: monitor disk
S ->> TS: start task server
A ->> TS: create task
TC ->> TS: subscribe task
TS ->> TC: distribute task
TC ->> CW: start worker
CW ->> TQ: add to task queue
TQ ->> CW: execute task
activate CW
CW ->> S: connect and auth
loop
SD ->> S: notify change
S ->> CW: notify change
CW ->> S: pull file
S ->> SD: read file
SD ->> S: return file
S ->> CW: send file
CW ->> CD: write file
end
deactivate CW
Monitor source directory and sync change files to dest directory.
You can use the logically_delete
flag to enable the logically delete and avoid deleting files by mistake.
Set the checkpoint_count
flag to use the checkpoint in the file to reduce transfer unmodified file chunks, by
default checkpoint_count=10
, which means it has 10+2
checkpoints at most. There are two additional checkpoints at
the head and tail. The first checkpoint is equal to the chunk_size
, it is optional. The last checkpoint is equal to
the file size, it is required. The checkpoint offset set by the checkpoint_count
is always more than chunk_size
,
unless the file size is less than or equal to chunk_size
, then the checkpoint_count
will be zero, so it is optional.
By default, if the file size and file modification time of the source file is equal to the destination file, then ignore
the current file transfer. You can use the force_checksum
flag to force enable the checksum to compare whether the
file is equal or not.
The default checksum hash algorithm is md5
, you can use the checksum_algorithm
flag to change the default hash
algorithm, current supported algorithms: md5
, sha1
, sha256
, sha512
, crc32
, crc64
, adler32
, fnv-1-32
, fnv-1a-32
, fnv-1-64
, fnv-1a-64
, fnv-1-128
, fnv-1a-128
.
If you want to reduce the frequency of synchronization, you can use the sync_delay
flag to enable sync delay, start
sync when the event count is equal or greater than sync_delay_events
, or wait for sync_delay_time
interval time
since the last sync.
And you can use the progress
flag to print the file sync progress bar.
$ gofs -source=./source -dest=./dest
You can use encrypt
flag to enable encryption and specify a directory as an encryption workspace by encrypt_path
flag. All files in the directory will be encrypted then sync to the destination path.
$ gofs -source=./source -dest=./dest -encrypt -encrypt_path=./source/encrypt -encrypt_secret=mysecret_16bytes
You can use the decrypt
flag to decrypt the encryption files to a specified path.
$ gofs -decrypt -decrypt_path=./dest/encrypt -decrypt_secret=mysecret_16bytes -decrypt_out=./decrypt_out
Sync the whole path immediately from source directory to dest directory.
$ gofs -source=./source -dest=./dest -sync_once
Sync the whole path from source directory to dest directory with cron.
# Per 30 seconds sync the whole path from source directory to dest directory
$ gofs -source=./source -dest=./dest -sync_cron="*/30 * * * * *"
Start a daemon to create subprocess to work, and record pid info to pid file.
$ gofs -source=./source -dest=./dest -daemon -daemon_pid
Start a file server for source directory and dest directory.
The file server is use HTTPS default, set the tls_cert_file
and tls_key_file
flags to customize the cert file and
key file.
You can disable the HTTPS by set the tls
flag to false
if you don't need it.
If you set the tls
to true
, the file server default port is 443
, otherwise it is 80
, and you can customize the
default port with the server_addr
flag, like -server_addr=":443"
.
If you enable the tls
flag on the server side, you can control whether a client skip verifies the server's certificate
chain and host name by the tls_insecure_skip_verify
flag, default is true
.
If you already enable the tls
flag, then you can use the http3
flag to enable the HTTP3 protocol in the server and
client sides.
You should set the rand_user_count
flag to auto generate some random users or set the users
flag to customize server
users for security reasons.
The server users will output to log if you set the rand_user_count
flag greater than zero.
If you need to compress the files, add the server_compress
flag to enable gzip compression for response, but it is not
fast now, and may reduce transmission efficiency in the LAN.
You can switch the session store mode for the file server by session_connection
flag,
currently supports memory and redis, default is memory.
If you want to use the redis as the session store, here is an example for redis session connection string:
redis://127.0.0.1:6379?password=redis_password&db=10&max_idle=10&secret=redis_secret
.
# Start a file server and create three random users
# Replace the `tls_cert_file` and `tls_key_file` flags with your real cert files in the production environment
$ gofs -source=./source -dest=./dest -server -tls_cert_file=cert.pem -tls_key_file=key.pem -rand_user_count=3
Use the max_tran_rate
flag to limit the max transmission rate in the server and client sides,
and this is an expected value, not an absolute one.
For example, limit the max transmission rate to 1048576 bytes, means 1MB.
$ gofs -source=./source -dest=./dest -max_tran_rate=1048576
Start a remote disk server as a remote file source.
The source
flag detail see Remote Server Source Protocol.
Pay attention to that remote disk server users must have read permission at least, for
example, -users="gofs|password|r"
.
You can use the checkpoint_count
and sync_delay
flags like the Local Disk.
# Start a remote disk server
# Replace the `tls_cert_file` and `tls_key_file` flags with your real cert files in the production environment
# Replace the `users` flag with complex username and password for security
$ gofs -source="rs://127.0.0.1:8105?mode=server&local_sync_disabled=true&path=./source&fs_server=https://127.0.0.1" -dest=./dest -users="gofs|password|r" -tls_cert_file=cert.pem -tls_key_file=key.pem -token_secret=mysecret_16bytes
Start a remote disk client to sync change files from remote disk server.
The source
flag detail see Remote Server Source Protocol.
Use the sync_once
flag to sync the whole path immediately from remote disk server to local dest directory,
like Sync Once.
Use the sync_cron
flag to sync the whole path from remote disk server to local dest directory with cron,
like Sync Cron.
Use the force_checksum
flag to force enable the checksum to compare whether the file is equal or not,
like Local Disk.
You can use the sync_delay
flag like the Local Disk.
# Start a remote disk client
# Replace the `users` flag with your real username and password
$ gofs -source="rs://127.0.0.1:8105" -dest=./dest -users="gofs|password" -tls_cert_file=cert.pem
Start a Remote Disk Server as a remote file source, then enable the remote push server with
the push_server
flag.
Pay attention to that remote push server users must have read and write permission at least, for
example, -users="gofs|password|rw"
.
# Start a remote disk server and enable the remote push server
# Replace the `tls_cert_file` and `tls_key_file` flags with your real cert files in the production environment
# Replace the `users` flag with complex username and password for security
$ gofs -source="rs://127.0.0.1:8105?mode=server&local_sync_disabled=true&path=./source&fs_server=https://127.0.0.1" -dest=./dest -users="gofs|password|rw" -tls_cert_file=cert.pem -tls_key_file=key.pem -push_server -token_secret=mysecret_16bytes
Start a remote push client to sync change files to the Remote Push Server.
Use the chunk_size
flag to set the chunk size of the big file to upload. The default value of chunk_size
is 1048576
, which means 1MB
.
You can use the checkpoint_count
and sync_delay
flags like the Local Disk.
More flag usage see Remote Disk Client.
# Start a remote push client and enable local disk sync, sync the file changes from source path to the local dest path and the remote push server
# Replace the `users` flag with your real username and password
$ gofs -source="./source" -dest="rs://127.0.0.1:8105?local_sync_disabled=false&path=./dest" -users="gofs|password" -tls_cert_file=cert.pem
Start a SFTP push client to sync change files to the SFTP server.
$ gofs -source="./source" -dest="sftp://127.0.0.1:22?local_sync_disabled=false&path=./dest&remote_path=/gofs_sftp_server&ssh_user=sftp_user&ssh_pass=sftp_pwd"
Start a SFTP pull client to pull the files from the SFTP server to the local destination path.
$ gofs -source="sftp://127.0.0.1:22?remote_path=/gofs_sftp_server&ssh_user=sftp_user&ssh_pass=sftp_pwd" -dest="./dest" -sync_once
Start a MinIO push client to sync change files to the MinIO server.
$ gofs -source="./source" -dest="minio://127.0.0.1:9000?secure=false&local_sync_disabled=false&path=./dest&remote_path=minio-bucket" -users="minio_user|minio_pwd"
Start a MinIO pull client to pull the files from the MinIO server to the local destination path.
$ gofs -source="minio://127.0.0.1:9000?secure=false&remote_path=minio-bucket" -dest="./dest" -users="minio_user|minio_pwd" -sync_once
Start a task server to distribute the tasks to clients.
Take the Remote Disk Server for example, create a task manifest config file like the remote-disk-task.yaml file first. Here defined a task that synchronizes files from server.
Then create the task content config file run-gofs-remote-disk-client.yaml that defined in the above manifest config file, and it will be executed by client.
Finally, start the remote disk server with the task_conf
flag.
Here use conf
to simplify the command and reuse the integration test config files.
$ cd integration
$ mkdir -p rs/source rs/dest
$ gofs -conf=./testdata/conf/run-gofs-remote-disk-server.yaml
Start a task client to subscribe to the task server, then acquire the task and execute it.
Use the task_client
flag to start the task client, and the task_client_max_worker
flag will limit the max
concurrent workers in the task client side.
And you can use the task_client_labels
flag to define the labels of the task client that use to match the task in the
task server side.
Here use conf
to simplify the command and reuse the integration test config files.
$ cd integration
$ mkdir -p rc/source rc/dest
$ gofs -conf=./testdata/conf/run-gofs-task-client.yaml
If you need to synchronize files between two devices that are unable to establish a direct connection, you can use a reverse proxy as a relay server. In more detail, see also Relay.
The remote server source protocol is based on URI, see RFC 3986.
The scheme name is rs
.
The remote server source uses 0.0.0.0
or other local ip address as host in Remote Disk Server
mode, and use ip address or domain name as host in Remote Disk Client mode.
The remote server source port, default is 8105
.
Use the following parameters in Remote Disk Server mode only.
path
the Remote Disk Server actual local source directorymode
running mode, in Remote Disk Server mode isserver
, default is running in Remote Disk Client modefs_server
File Server address, likehttps://127.0.0.1
local_sync_disabled
disabled Remote Disk Server sync changes to its local dest path,true
orfalse
, default isfalse
For example, in Remote Disk Server mode.
rs://127.0.0.1:8105?mode=server&local_sync_disabled=true&path=./source&fs_server=https://127.0.0.1
\_/ \_______/ \__/ \____________________________________________________________________________/
| | | |
scheme host port parameter
Enable manage api base on File Server by using the manage
flag.
By default, allow to access manage api by private address and loopback address only.
You can disable it by setting the manage_private
flag to false
.
$ gofs -source=./source -dest=./dest -server -tls_cert_file=cert.pem -tls_key_file=key.pem -rand_user_count=3 -manage
The pprof url address like this
https://127.0.0.1/manage/pprof/
Reading the program config, default return the config with json
format, and support json
and yaml
format
currently.
https://127.0.0.1/manage/config
Or use the format
parameter to specific the config format.
https://127.0.0.1/manage/config?format=yaml
Use the report
flag to enable report api route, and start to collect the report data, need to enable the manage
flag
first.
The details of the report api see Report API.
https://127.0.0.1/manage/report
Enable the file logger and console logger by default, and you can disable the file logger by setting the log_file
flag
to false
.
Use the log_level
flag to set the log level, default is INFO
, (DEBUG=0
INFO=1
WARN=2
ERROR=3
).
Use the log_dir
flag to set the directory of the log file, default is ./logs/
.
Use the log_flush
flag to enable auto flush log with interval, default is true
.
Use the log_flush_interval
flag to set the log flush interval duration, default is 3s
.
Use the log_event
flag to enable the event log, write to file, default is false
.
Use the log_sample_rate
flag to set the sample rate for the sample logger, and the value ranges from 0 to 1, default
is 1
.
Use the log_format
flag to set the log output format, current support text
and json
, default is text
.
Use the log_split_date
flag to split log file by date, default is false
.
# set the logger config in "Local Disk" mode
$ gofs -source=./source -dest=./dest -log_file -log_level=0 -log_dir="./logs/" -log_flush -log_flush_interval=3s -log_event
If you want, you can use a configuration file to replace all the flags.It supports json
and yaml
format currently.
All the configuration fields are the same as the flags, you can refer to the Configuration Example or the response of Config API.
$ gofs -conf=./gofs.yaml
You can use the checksum
flag to calculate the file checksum and print the result.
The chunk_size
, checkpoint_count
and checksum_algorithm
flags are effective here the same as in
the Local Disk.
$ gofs -source=./gofs -checksum
$ gofs -h
$ gofs -v
$ gofs -about
The gofs-webui is a web UI tool for gofs
, and it allows you to generate the
config file of gofs
through the web UI, making the use of gofs easier.
%%{init: { "flowchart": {"htmlLabels": false}} }%%
flowchart TD
PR[pull request]
MainRepo[github.com/no-src/gofs.git] -- 1.fork --> ForkRepo[github.com/yourname/gofs.git]
ForkRepo -- 2.git clone --> LocalRepo[local repository]
LocalRepo -- 3.commit changes --> NewBranch[new branch]
NewBranch -- 4.git push --> ForkRepo
ForkRepo -- 5.create pull request --> PR
PR -- 6.merge to --> MainRepo
Quick to contribute using cloud development environments.