garage/src/main.rs

458 lines
10 KiB
Rust
Raw Normal View History

#![recursion_limit = "1024"]
2020-04-21 12:54:55 +00:00
#[macro_use]
extern crate log;
mod data;
mod error;
2020-04-08 20:00:41 +00:00
mod background;
2020-04-06 17:55:39 +00:00
mod membership;
2020-04-08 20:00:41 +00:00
mod table;
mod table_fullcopy;
mod table_sharded;
2020-04-16 12:50:49 +00:00
mod table_sync;
2020-04-08 20:00:41 +00:00
mod block;
2020-04-10 21:11:52 +00:00
mod block_ref_table;
mod bucket_table;
2020-04-09 15:32:28 +00:00
mod object_table;
2020-04-09 21:45:07 +00:00
mod version_table;
2020-04-08 20:00:41 +00:00
mod admin_rpc;
mod api_server;
2020-04-10 20:26:48 +00:00
mod http_util;
mod rpc_client;
mod rpc_server;
mod server;
2020-04-12 13:51:19 +00:00
mod tls_util;
use serde::{Deserialize, Serialize};
use std::collections::HashSet;
2020-04-06 17:55:39 +00:00
use std::net::SocketAddr;
use std::path::PathBuf;
2020-04-18 17:21:34 +00:00
use std::sync::Arc;
use std::time::Duration;
use structopt::StructOpt;
use error::Error;
use membership::*;
2020-04-18 17:21:34 +00:00
use rpc_client::*;
use server::TlsConfig;
use admin_rpc::*;
#[derive(StructOpt, Debug)]
#[structopt(name = "garage")]
pub struct Opt {
/// RPC connect to this host to execute client operations
#[structopt(short = "h", long = "rpc-host", default_value = "127.0.0.1:3901")]
rpc_host: SocketAddr,
2020-04-06 17:55:39 +00:00
2020-04-16 12:50:49 +00:00
#[structopt(long = "ca-cert")]
2020-04-12 17:41:19 +00:00
ca_cert: Option<String>,
2020-04-16 12:50:49 +00:00
#[structopt(long = "client-cert")]
2020-04-12 17:41:19 +00:00
client_cert: Option<String>,
2020-04-16 12:50:49 +00:00
#[structopt(long = "client-key")]
2020-04-12 17:41:19 +00:00
client_key: Option<String>,
#[structopt(subcommand)]
cmd: Command,
}
#[derive(StructOpt, Debug)]
pub enum Command {
/// Run Garage server
#[structopt(name = "server")]
Server(ServerOpt),
/// Get network status
#[structopt(name = "status")]
Status,
/// Garage node operations
#[structopt(name = "node")]
Node(NodeOperation),
2020-04-16 15:04:28 +00:00
/// Bucket operations
#[structopt(name = "bucket")]
Bucket(BucketOperation),
2020-04-19 20:36:36 +00:00
/// Start repair of node data
#[structopt(name = "repair")]
Repair(RepairOpt),
2020-04-06 17:55:39 +00:00
}
#[derive(StructOpt, Debug)]
pub struct ServerOpt {
/// Configuration file
#[structopt(short = "c", long = "config", default_value = "./config.toml")]
config_file: PathBuf,
}
#[derive(StructOpt, Debug)]
pub enum NodeOperation {
/// Configure Garage node
#[structopt(name = "configure")]
Configure(ConfigureNodeOpt),
/// Remove Garage node from cluster
#[structopt(name = "remove")]
Remove(RemoveNodeOpt),
}
#[derive(StructOpt, Debug)]
pub struct ConfigureNodeOpt {
/// Node to configure (prefix of hexadecimal node id)
node_id: String,
2020-04-07 15:00:48 +00:00
/// Location (datacenter) of the node
datacenter: String,
/// Number of tokens
n_tokens: u32,
2020-04-21 14:07:15 +00:00
/// Optionnal node tag
#[structopt(long = "tag", default_value = "")]
tag: String,
}
2020-04-16 15:04:28 +00:00
#[derive(StructOpt, Debug)]
pub struct RemoveNodeOpt {
2020-04-16 15:04:28 +00:00
/// Node to configure (prefix of hexadecimal node id)
node_id: String,
/// If this flag is not given, the node won't be removed
#[structopt(long = "yes")]
yes: bool,
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub enum BucketOperation {
/// List buckets
#[structopt(name = "list")]
List,
/// Get bucket info
#[structopt(name = "info")]
Info(BucketOpt),
/// Create bucket
#[structopt(name = "create")]
Create(BucketOpt),
/// Delete bucket
#[structopt(name = "delete")]
Delete(DeleteBucketOpt),
/// Allow key to read or write to bucket
#[structopt(name = "allow")]
Allow(PermBucketOpt),
/// Allow key to read or write to bucket
#[structopt(name = "deny")]
Deny(PermBucketOpt),
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub struct BucketOpt {
/// Bucket name
pub name: String,
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub struct DeleteBucketOpt {
/// Bucket name
pub name: String,
/// If this flag is not given, the bucket won't be deleted
#[structopt(long = "yes")]
pub yes: bool,
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub struct PermBucketOpt {
/// Access key ID
#[structopt(long = "key")]
pub key: String,
/// Allow/deny read operations
#[structopt(long = "read")]
pub read: bool,
/// Allow/deny write operations
#[structopt(long = "write")]
pub write: bool,
/// Bucket name
pub bucket: String,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Clone)]
2020-04-19 20:36:36 +00:00
pub struct RepairOpt {
/// Launch repair operation on all nodes
#[structopt(short = "a", long = "all-nodes")]
pub all_nodes: bool,
/// Confirm the launch of the repair operation
#[structopt(long = "yes")]
pub yes: bool,
#[structopt(subcommand)]
pub what: Option<RepairWhat>,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum RepairWhat {
/// Only do a full sync of metadata tables
#[structopt(name = "tables")]
Tables,
/// Only repair (resync/rebalance) the set of stored blocks
#[structopt(name = "blocks")]
Blocks,
/// Only redo the propagation of object deletions to the version table (slow)
#[structopt(name = "versions")]
Versions,
/// Only redo the propagation of version deletions to the block ref table (extremely slow)
#[structopt(name = "block_refs")]
BlockRefs,
2020-04-19 20:36:36 +00:00
}
#[tokio::main]
async fn main() {
2020-04-21 12:54:55 +00:00
pretty_env_logger::init();
let opt = Opt::from_args();
2020-04-12 17:41:19 +00:00
let tls_config = match (opt.ca_cert, opt.client_cert, opt.client_key) {
2020-04-16 12:50:49 +00:00
(Some(ca_cert), Some(client_cert), Some(client_key)) => Some(TlsConfig {
ca_cert,
node_cert: client_cert,
node_key: client_key,
}),
2020-04-12 17:41:19 +00:00
(None, None, None) => None,
_ => {
2020-04-21 12:54:55 +00:00
warn!("Missing one of: --ca-cert, --node-cert, --node-key. Not using TLS.");
2020-04-12 17:41:19 +00:00
None
}
};
2020-04-18 17:21:34 +00:00
let rpc_http_cli =
Arc::new(RpcHttpClient::new(&tls_config).expect("Could not create RPC client"));
let membership_rpc_cli =
RpcAddrClient::new(rpc_http_cli.clone(), MEMBERSHIP_RPC_PATH.to_string());
let admin_rpc_cli = RpcAddrClient::new(rpc_http_cli.clone(), ADMIN_RPC_PATH.to_string());
let resp = match opt.cmd {
2020-04-16 15:04:28 +00:00
Command::Server(server_opt) => {
// Abort on panic (same behavior as in Go)
std::panic::set_hook(Box::new(|panic_info| {
2020-04-21 12:54:55 +00:00
error!("{}", panic_info.to_string());
2020-04-16 15:04:28 +00:00
std::process::abort();
}));
server::run_server(server_opt.config_file).await
}
Command::Status => cmd_status(membership_rpc_cli, opt.rpc_host).await,
Command::Node(NodeOperation::Configure(configure_opt)) => {
cmd_configure(membership_rpc_cli, opt.rpc_host, configure_opt).await
}
Command::Node(NodeOperation::Remove(remove_opt)) => {
cmd_remove(membership_rpc_cli, opt.rpc_host, remove_opt).await
}
Command::Bucket(bo) => {
cmd_admin(admin_rpc_cli, opt.rpc_host, AdminRPC::BucketOperation(bo)).await
}
2020-04-19 20:36:36 +00:00
Command::Repair(ro) => {
cmd_admin(admin_rpc_cli, opt.rpc_host, AdminRPC::LaunchRepair(ro)).await
2020-04-19 20:36:36 +00:00
}
};
if let Err(e) = resp {
2020-04-21 12:54:55 +00:00
error!("Error: {}", e);
}
}
2020-04-18 17:21:34 +00:00
async fn cmd_status(rpc_cli: RpcAddrClient<Message>, rpc_host: SocketAddr) -> Result<(), Error> {
let status = match rpc_cli
2020-04-19 20:36:36 +00:00
.call(&rpc_host, &Message::PullStatus, ADMIN_RPC_TIMEOUT)
.await?
{
Message::AdvertiseNodesUp(nodes) => nodes,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
let config = match rpc_cli
2020-04-19 20:36:36 +00:00
.call(&rpc_host, &Message::PullConfig, ADMIN_RPC_TIMEOUT)
.await?
{
Message::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
println!("Healthy nodes:");
for adv in status.iter() {
if let Some(cfg) = config.members.get(&adv.id) {
println!(
2020-04-21 14:07:15 +00:00
"{:?}\t{}\t{}\t{}\t{}\t{}",
adv.id, adv.state_info.hostname, adv.addr, cfg.tag, cfg.datacenter, cfg.n_tokens
);
}
}
let status_keys = status.iter().map(|x| x.id).collect::<HashSet<_>>();
if config
.members
.iter()
.any(|(id, _)| !status_keys.contains(id))
{
println!("\nFailed nodes:");
for (id, cfg) in config.members.iter() {
if !status.iter().any(|x| x.id == *id) {
2020-04-21 14:07:15 +00:00
println!(
"{:?}\t{}\t{}\t{}",
id, cfg.tag, cfg.datacenter, cfg.n_tokens
);
}
}
}
if status
.iter()
.any(|adv| !config.members.contains_key(&adv.id))
{
println!("\nUnconfigured nodes:");
for adv in status.iter() {
if !config.members.contains_key(&adv.id) {
2020-04-19 17:08:48 +00:00
println!("{:?}\t{}\t{}", adv.id, adv.state_info.hostname, adv.addr);
}
}
}
Ok(())
}
2020-04-06 17:55:39 +00:00
async fn cmd_configure(
2020-04-18 17:21:34 +00:00
rpc_cli: RpcAddrClient<Message>,
rpc_host: SocketAddr,
args: ConfigureNodeOpt,
) -> Result<(), Error> {
let status = match rpc_cli
2020-04-19 20:36:36 +00:00
.call(&rpc_host, &Message::PullStatus, ADMIN_RPC_TIMEOUT)
.await?
{
Message::AdvertiseNodesUp(nodes) => nodes,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
let mut candidates = vec![];
for adv in status.iter() {
if hex::encode(&adv.id).starts_with(&args.node_id) {
candidates.push(adv.id);
}
}
if candidates.len() != 1 {
return Err(Error::Message(format!(
"{} matching nodes",
candidates.len()
)));
}
let mut config = match rpc_cli
2020-04-19 20:36:36 +00:00
.call(&rpc_host, &Message::PullConfig, ADMIN_RPC_TIMEOUT)
.await?
{
Message::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
config.members.insert(
candidates[0].clone(),
NetworkConfigEntry {
datacenter: args.datacenter,
n_tokens: args.n_tokens,
2020-04-21 14:07:15 +00:00
tag: args.tag,
},
);
config.version += 1;
rpc_cli
.call(
&rpc_host,
&Message::AdvertiseConfig(config),
2020-04-19 20:36:36 +00:00
ADMIN_RPC_TIMEOUT,
)
.await?;
Ok(())
}
2020-04-16 15:04:28 +00:00
async fn cmd_remove(
2020-04-18 17:21:34 +00:00
rpc_cli: RpcAddrClient<Message>,
2020-04-16 15:04:28 +00:00
rpc_host: SocketAddr,
args: RemoveNodeOpt,
2020-04-16 15:04:28 +00:00
) -> Result<(), Error> {
let mut config = match rpc_cli
2020-04-19 20:36:36 +00:00
.call(&rpc_host, &Message::PullConfig, ADMIN_RPC_TIMEOUT)
2020-04-16 15:04:28 +00:00
.await?
{
Message::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
let mut candidates = vec![];
for (key, _) in config.members.iter() {
if hex::encode(key).starts_with(&args.node_id) {
candidates.push(*key);
2020-04-16 15:04:28 +00:00
}
}
if candidates.len() != 1 {
return Err(Error::Message(format!(
"{} matching nodes",
candidates.len()
)));
}
if !args.yes {
return Err(Error::Message(format!(
"Add the flag --yes to really remove {:?} from the cluster",
candidates[0]
)));
}
config.members.remove(&candidates[0]);
config.version += 1;
rpc_cli
.call(
&rpc_host,
&Message::AdvertiseConfig(config),
2020-04-19 20:36:36 +00:00
ADMIN_RPC_TIMEOUT,
2020-04-16 15:04:28 +00:00
)
.await?;
Ok(())
}
async fn cmd_admin(
rpc_cli: RpcAddrClient<AdminRPC>,
rpc_host: SocketAddr,
args: AdminRPC,
) -> Result<(), Error> {
2020-04-19 20:36:36 +00:00
match rpc_cli.call(&rpc_host, args, ADMIN_RPC_TIMEOUT).await? {
AdminRPC::Ok(msg) => {
println!("{}", msg);
}
AdminRPC::BucketList(bl) => {
println!("List of buckets:");
for bucket in bl {
println!("{}", bucket);
}
}
AdminRPC::BucketInfo(bucket) => {
println!("{:?}", bucket);
}
r => {
2020-04-21 12:54:55 +00:00
error!("Unexpected response: {:?}", r);
}
}
Ok(())
}