Improvements to CLI and various fixes for netapp version

Discovery via consul, persist peer list to file
This commit is contained in:
Alex 2021-10-15 11:05:09 +02:00
parent 4067797d01
commit 1b450c4b49
No known key found for this signature in database
GPG key ID: EDABF9711E244EB1
26 changed files with 1415 additions and 1523 deletions

225
Cargo.lock generated
View file

@ -83,12 +83,6 @@ dependencies = [
"generic-array", "generic-array",
] ]
[[package]]
name = "bumpalo"
version = "3.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d9df67f7bf9ef8498769f994239c45613ef0c5899415fb58e9add412d2c1a538"
[[package]] [[package]]
name = "byteorder" name = "byteorder"
version = "1.4.3" version = "1.4.3"
@ -393,9 +387,9 @@ dependencies = [
"futures-util", "futures-util",
"garage_api", "garage_api",
"garage_model", "garage_model",
"garage_rpc 0.3.0", "garage_rpc",
"garage_table", "garage_table",
"garage_util 0.3.0", "garage_util",
"garage_web", "garage_web",
"git-version", "git-version",
"hex", "hex",
@ -425,7 +419,7 @@ dependencies = [
"futures-util", "futures-util",
"garage_model", "garage_model",
"garage_table", "garage_table",
"garage_util 0.3.0", "garage_util",
"hex", "hex",
"hmac", "hmac",
"http", "http",
@ -451,9 +445,9 @@ dependencies = [
"async-trait", "async-trait",
"futures", "futures",
"futures-util", "futures-util",
"garage_rpc 0.3.0", "garage_rpc",
"garage_table", "garage_table",
"garage_util 0.3.0", "garage_util",
"hex", "hex",
"log", "log",
"netapp", "netapp",
@ -474,8 +468,7 @@ dependencies = [
"bytes 1.1.0", "bytes 1.1.0",
"futures", "futures",
"futures-util", "futures-util",
"garage_rpc 0.2.1", "garage_util",
"garage_util 0.3.0",
"gethostname", "gethostname",
"hex", "hex",
"hyper", "hyper",
@ -498,8 +491,8 @@ dependencies = [
"bytes 1.1.0", "bytes 1.1.0",
"futures", "futures",
"futures-util", "futures-util",
"garage_rpc 0.3.0", "garage_rpc",
"garage_util 0.3.0", "garage_util",
"hexdump", "hexdump",
"log", "log",
"rand", "rand",
@ -543,7 +536,7 @@ dependencies = [
"garage_api", "garage_api",
"garage_model", "garage_model",
"garage_table", "garage_table",
"garage_util 0.3.0", "garage_util",
"http", "http",
"hyper", "hyper",
"idna", "idna",
@ -604,31 +597,6 @@ dependencies = [
"syn", "syn",
] ]
[[package]]
name = "h2"
version = "0.3.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c06815895acec637cd6ed6e9662c935b866d20a106f8361892893a7d9234964"
dependencies = [
"bytes 1.1.0",
"fnv",
"futures-core",
"futures-sink",
"futures-util",
"http",
"indexmap",
"slab",
"tokio",
"tokio-util",
"tracing",
]
[[package]]
name = "hashbrown"
version = "0.11.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ab5ef0d4909ef3724cc8cce6ccc8572c5c817592e9285f5464f8e86f8bd3726e"
[[package]] [[package]]
name = "heck" name = "heck"
version = "0.3.3" version = "0.3.3"
@ -738,7 +706,6 @@ dependencies = [
"futures-channel", "futures-channel",
"futures-core", "futures-core",
"futures-util", "futures-util",
"h2",
"http", "http",
"http-body", "http-body",
"httparse", "httparse",
@ -752,21 +719,6 @@ dependencies = [
"want", "want",
] ]
[[package]]
name = "hyper-rustls"
version = "0.22.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5f9f7a97316d44c0af9b0301e65010573a853a9fc97046d7331d7f6bc0fd5a64"
dependencies = [
"futures-util",
"hyper",
"log",
"rustls",
"tokio",
"tokio-rustls",
"webpki",
]
[[package]] [[package]]
name = "idna" name = "idna"
version = "0.2.3" version = "0.2.3"
@ -778,16 +730,6 @@ dependencies = [
"unicode-normalization", "unicode-normalization",
] ]
[[package]]
name = "indexmap"
version = "1.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bc633605454125dec4b66843673f01c7df2b89479b32e0ed634e43a91cff62a5"
dependencies = [
"autocfg",
"hashbrown",
]
[[package]] [[package]]
name = "instant" name = "instant"
version = "0.1.11" version = "0.1.11"
@ -809,19 +751,11 @@ version = "0.4.8"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b71991ff56294aa922b450139ee08b3bfc70982c6b2c7562771375cf73542dd4" checksum = "b71991ff56294aa922b450139ee08b3bfc70982c6b2c7562771375cf73542dd4"
[[package]]
name = "js-sys"
version = "0.3.55"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7cc9ffccd38c451a86bf13657df244e9c3f37493cce8e5e21e940963777acc84"
dependencies = [
"wasm-bindgen",
]
[[package]] [[package]]
name = "kuska-handshake" name = "kuska-handshake"
version = "0.2.0" version = "0.2.0"
source = "git+https://github.com/Alexis211/handshake?branch=tokio1.0#61bf144643b177797b2d16b9b2ffcfb648face00" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e33da4b69f23c2ece0b3e729d079cebdc2c0206e493e42f510f500ad81c631d5"
dependencies = [ dependencies = [
"futures", "futures",
"hex", "hex",
@ -940,7 +874,7 @@ dependencies = [
[[package]] [[package]]
name = "netapp" name = "netapp"
version = "0.3.0" version = "0.3.0"
source = "git+https://git.deuxfleurs.fr/lx/netapp#cfa64bc745969cfc3684a70b45d71128f8335460" source = "git+https://git.deuxfleurs.fr/lx/netapp#9b64c27da68f7ac9049e02e26da918e871a63f07"
dependencies = [ dependencies = [
"arc-swap", "arc-swap",
"async-trait", "async-trait",
@ -1209,21 +1143,6 @@ version = "0.6.25"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f497285884f3fcff424ffc933e56d7cbca511def0c9831a7f9b5f6153e3cc89b" checksum = "f497285884f3fcff424ffc933e56d7cbca511def0c9831a7f9b5f6153e3cc89b"
[[package]]
name = "ring"
version = "0.16.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3053cf52e236a3ed746dfc745aa9cacf1b791d846bdaf412f60a8d7d6e17c8fc"
dependencies = [
"cc",
"libc",
"once_cell",
"spin",
"untrusted",
"web-sys",
"winapi",
]
[[package]] [[package]]
name = "rmp" name = "rmp"
version = "0.8.10" version = "0.8.10"
@ -1265,19 +1184,6 @@ dependencies = [
"xmlparser", "xmlparser",
] ]
[[package]]
name = "rustls"
version = "0.19.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "35edb675feee39aec9c99fa5ff985081995a06d594114ae14cbe797ad7b7a6d7"
dependencies = [
"base64",
"log",
"ring",
"sct",
"webpki",
]
[[package]] [[package]]
name = "rustversion" name = "rustversion"
version = "1.0.5" version = "1.0.5"
@ -1305,16 +1211,6 @@ version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd" checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd"
[[package]]
name = "sct"
version = "0.6.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b362b83898e0e69f38515b82ee15aa80636befe47c3b6d3d89a911e78fc228ce"
dependencies = [
"ring",
"untrusted",
]
[[package]] [[package]]
name = "serde" name = "serde"
version = "1.0.130" version = "1.0.130"
@ -1415,12 +1311,6 @@ dependencies = [
"winapi", "winapi",
] ]
[[package]]
name = "spin"
version = "0.5.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
[[package]] [[package]]
name = "structopt" name = "structopt"
version = "0.3.23" version = "0.3.23"
@ -1567,17 +1457,6 @@ dependencies = [
"syn", "syn",
] ]
[[package]]
name = "tokio-rustls"
version = "0.22.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bc6844de72e57df1980054b38be3a9f4702aba4858be64dd700181a8a6d0e1b6"
dependencies = [
"rustls",
"tokio",
"webpki",
]
[[package]] [[package]]
name = "tokio-stream" name = "tokio-stream"
version = "0.1.7" version = "0.1.7"
@ -1684,12 +1563,6 @@ version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3" checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3"
[[package]]
name = "untrusted"
version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a156c684c91ea7d62626509bce3cb4e1d9ed5c4d978f7b4352658f96a4c26b4a"
[[package]] [[package]]
name = "url" name = "url"
version = "2.2.2" version = "2.2.2"
@ -1735,80 +1608,6 @@ version = "0.10.2+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6" checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6"
[[package]]
name = "wasm-bindgen"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "632f73e236b219150ea279196e54e610f5dbafa5d61786303d4da54f84e47fce"
dependencies = [
"cfg-if",
"wasm-bindgen-macro",
]
[[package]]
name = "wasm-bindgen-backend"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a317bf8f9fba2476b4b2c85ef4c4af8ff39c3c7f0cdfeed4f82c34a880aa837b"
dependencies = [
"bumpalo",
"lazy_static",
"log",
"proc-macro2",
"quote",
"syn",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-macro"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d56146e7c495528bf6587663bea13a8eb588d39b36b679d83972e1a2dbbdacf9"
dependencies = [
"quote",
"wasm-bindgen-macro-support",
]
[[package]]
name = "wasm-bindgen-macro-support"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7803e0eea25835f8abdc585cd3021b3deb11543c6fe226dcd30b228857c5c5ab"
dependencies = [
"proc-macro2",
"quote",
"syn",
"wasm-bindgen-backend",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-shared"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0237232789cf037d5480773fe568aac745bfe2afbc11a863e97901780a6b47cc"
[[package]]
name = "web-sys"
version = "0.3.55"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "38eb105f1c59d9eaa6b5cdc92b859d85b926e82cb2e0945cd0c9259faa6fe9fb"
dependencies = [
"js-sys",
"wasm-bindgen",
]
[[package]]
name = "webpki"
version = "0.21.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b8e38c0608262c46d4a56202ebabdeb094cef7e560ca7a226c6bf055188aa4ea"
dependencies = [
"ring",
"untrusted",
]
[[package]] [[package]]
name = "winapi" name = "winapi"
version = "0.3.9" version = "0.3.9"

1635
Cargo.nix

File diff suppressed because it is too large Load diff

View file

@ -35,7 +35,7 @@ tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi
http = "0.2" http = "0.2"
httpdate = "0.3" httpdate = "0.3"
http-range = "0.1" http-range = "0.1"
hyper = "0.14" hyper = { version = "0.14", features = ["server", "http1", "runtime", "tcp", "stream"] }
percent-encoding = "2.1.0" percent-encoding = "2.1.0"
roxmltree = "0.14" roxmltree = "0.14"
serde = { version = "1.0", features = ["derive"] } serde = { version = "1.0", features = ["derive"] }

View file

@ -82,7 +82,9 @@ impl Error {
match self { match self {
Error::NotFound => StatusCode::NOT_FOUND, Error::NotFound => StatusCode::NOT_FOUND,
Error::Forbidden(_) => StatusCode::FORBIDDEN, Error::Forbidden(_) => StatusCode::FORBIDDEN,
Error::InternalError(GarageError::Rpc(_)) => StatusCode::SERVICE_UNAVAILABLE, Error::InternalError(
GarageError::Timeout | GarageError::RemoteError(_) | GarageError::TooManyErrors(_),
) => StatusCode::SERVICE_UNAVAILABLE,
Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => { Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => {
StatusCode::INTERNAL_SERVER_ERROR StatusCode::INTERNAL_SERVER_ERROR
} }
@ -95,7 +97,9 @@ impl Error {
Error::NotFound => "NoSuchKey", Error::NotFound => "NoSuchKey",
Error::Forbidden(_) => "AccessDenied", Error::Forbidden(_) => "AccessDenied",
Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed", Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed",
Error::InternalError(GarageError::Rpc(_)) => "ServiceUnavailable", Error::InternalError(
GarageError::Timeout | GarageError::RemoteError(_) | GarageError::TooManyErrors(_),
) => "ServiceUnavailable",
Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => "InternalError", Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => "InternalError",
_ => "InvalidRequest", _ => "InvalidRequest",
} }

View file

@ -31,15 +31,14 @@ pub enum AdminRpc {
// Replies // Replies
Ok(String), Ok(String),
Error(String),
BucketList(Vec<String>), BucketList(Vec<String>),
BucketInfo(Bucket), BucketInfo(Bucket),
KeyList(Vec<(String, String)>), KeyList(Vec<(String, String)>),
KeyInfo(Key), KeyInfo(Key),
} }
impl Message for AdminRpc { impl Rpc for AdminRpc {
type Response = AdminRpc; type Response = Result<AdminRpc, Error>;
} }
pub struct AdminRpcHandler { pub struct AdminRpcHandler {
@ -341,17 +340,20 @@ impl AdminRpcHandler {
let mut failures = vec![]; let mut failures = vec![];
let ring = self.garage.system.ring.borrow().clone(); let ring = self.garage.system.ring.borrow().clone();
for node in ring.config.members.keys() { for node in ring.config.members.keys() {
let node = NodeID::from_slice(node.as_slice()).unwrap(); let node = (*node).into();
if self let resp = self
.endpoint .endpoint
.call( .call(
&node, &node,
&AdminRpc::LaunchRepair(opt_to_send.clone()), &AdminRpc::LaunchRepair(opt_to_send.clone()),
PRIO_NORMAL, PRIO_NORMAL,
) )
.await .await;
.is_err() let is_err = match resp {
{ Ok(Ok(_)) => false,
_ => true,
};
if is_err {
failures.push(node); failures.push(node);
} }
} }
@ -386,17 +388,17 @@ impl AdminRpcHandler {
let ring = self.garage.system.ring.borrow().clone(); let ring = self.garage.system.ring.borrow().clone();
for node in ring.config.members.keys() { for node in ring.config.members.keys() {
let node = NodeID::from_slice(node.as_slice()).unwrap();
let mut opt = opt.clone(); let mut opt = opt.clone();
opt.all_nodes = false; opt.all_nodes = false;
writeln!(&mut ret, "\n======================").unwrap(); writeln!(&mut ret, "\n======================").unwrap();
writeln!(&mut ret, "Stats for node {:?}:", node).unwrap(); writeln!(&mut ret, "Stats for node {:?}:", node).unwrap();
let node_id = (*node).into();
match self match self
.endpoint .endpoint
.call(&node, &AdminRpc::Stats(opt), PRIO_NORMAL) .call(&node_id, &AdminRpc::Stats(opt), PRIO_NORMAL)
.await .await?
{ {
Ok(AdminRpc::Ok(s)) => writeln!(&mut ret, "{}", s).unwrap(), Ok(AdminRpc::Ok(s)) => writeln!(&mut ret, "{}", s).unwrap(),
Ok(x) => writeln!(&mut ret, "Bad answer: {:?}", x).unwrap(), Ok(x) => writeln!(&mut ret, "Bad answer: {:?}", x).unwrap(),
@ -486,9 +488,16 @@ impl AdminRpcHandler {
.unwrap(); .unwrap();
writeln!(to, " GC todo queue length: {}", t.data.gc_todo_len()).unwrap(); writeln!(to, " GC todo queue length: {}", t.data.gc_todo_len()).unwrap();
} }
}
async fn handle_rpc(self: &Arc<Self>, msg: &AdminRpc) -> Result<AdminRpc, Error> { #[async_trait]
match msg { impl EndpointHandler<AdminRpc> for AdminRpcHandler {
async fn handle(
self: &Arc<Self>,
message: &AdminRpc,
_from: NodeID,
) -> Result<AdminRpc, Error> {
match message {
AdminRpc::BucketOperation(bo) => self.handle_bucket_cmd(bo).await, AdminRpc::BucketOperation(bo) => self.handle_bucket_cmd(bo).await,
AdminRpc::KeyOperation(ko) => self.handle_key_cmd(ko).await, AdminRpc::KeyOperation(ko) => self.handle_key_cmd(ko).await,
AdminRpc::LaunchRepair(opt) => self.handle_launch_repair(opt.clone()).await, AdminRpc::LaunchRepair(opt) => self.handle_launch_repair(opt.clone()).await,
@ -497,12 +506,3 @@ impl AdminRpcHandler {
} }
} }
} }
#[async_trait]
impl EndpointHandler<AdminRpc> for AdminRpcHandler {
async fn handle(self: &Arc<Self>, message: &AdminRpc, _from: NodeID) -> AdminRpc {
self.handle_rpc(message)
.await
.unwrap_or_else(|e| AdminRpc::Error(format!("{}", e)))
}
}

View file

@ -1,5 +1,4 @@
//use std::cmp::max; use std::collections::HashSet;
//use std::collections::HashSet;
use std::path::PathBuf; use std::path::PathBuf;
use serde::{Deserialize, Serialize}; use serde::{Deserialize, Serialize};
@ -7,7 +6,7 @@ use structopt::StructOpt;
use garage_util::data::Uuid; use garage_util::data::Uuid;
use garage_util::error::Error; use garage_util::error::Error;
//use garage_util::time::*; use garage_util::time::*;
use garage_rpc::ring::*; use garage_rpc::ring::*;
use garage_rpc::system::*; use garage_rpc::system::*;
@ -58,6 +57,10 @@ pub struct ServerOpt {
#[derive(StructOpt, Debug)] #[derive(StructOpt, Debug)]
pub enum NodeOperation { pub enum NodeOperation {
/// Connect to Garage node that is currently isolated from the system
#[structopt(name = "connect")]
Connect(ConnectNodeOpt),
/// Configure Garage node /// Configure Garage node
#[structopt(name = "configure")] #[structopt(name = "configure")]
Configure(ConfigureNodeOpt), Configure(ConfigureNodeOpt),
@ -67,6 +70,13 @@ pub enum NodeOperation {
Remove(RemoveNodeOpt), Remove(RemoveNodeOpt),
} }
#[derive(StructOpt, Debug)]
pub struct ConnectNodeOpt {
/// Node public key and address, in the format:
/// `<public key hexadecimal>@<ip or hostname>:<port>`
node: String,
}
#[derive(StructOpt, Debug)] #[derive(StructOpt, Debug)]
pub struct ConfigureNodeOpt { pub struct ConfigureNodeOpt {
/// Node to configure (prefix of hexadecimal node id) /// Node to configure (prefix of hexadecimal node id)
@ -303,6 +313,9 @@ pub async fn cli_cmd(
) -> Result<(), Error> { ) -> Result<(), Error> {
match cmd { match cmd {
Command::Status => cmd_status(system_rpc_endpoint, rpc_host).await, Command::Status => cmd_status(system_rpc_endpoint, rpc_host).await,
Command::Node(NodeOperation::Connect(connect_opt)) => {
cmd_connect(system_rpc_endpoint, rpc_host, connect_opt).await
}
Command::Node(NodeOperation::Configure(configure_opt)) => { Command::Node(NodeOperation::Configure(configure_opt)) => {
cmd_configure(system_rpc_endpoint, rpc_host, configure_opt).await cmd_configure(system_rpc_endpoint, rpc_host, configure_opt).await
} }
@ -326,142 +339,96 @@ pub async fn cli_cmd(
pub async fn cmd_status(rpc_cli: &Endpoint<SystemRpc, ()>, rpc_host: NodeID) -> Result<(), Error> { pub async fn cmd_status(rpc_cli: &Endpoint<SystemRpc, ()>, rpc_host: NodeID) -> Result<(), Error> {
let status = match rpc_cli let status = match rpc_cli
.call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL) .call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL)
.await? .await??
{ {
SystemRpc::ReturnKnownNodes(nodes) => nodes, SystemRpc::ReturnKnownNodes(nodes) => nodes,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))), resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
}; };
let config = match rpc_cli let config = match rpc_cli
.call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL) .call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL)
.await? .await??
{ {
SystemRpc::AdvertiseConfig(cfg) => cfg, SystemRpc::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))), resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
}; };
println!("STATUS:");
for node in status {
println!("{:?}", node);
}
println!("CONFIG: (v{})", config.version);
for (id, node) in config.members {
println!("{} {:?}", hex::encode(id.as_slice()), node);
}
/* TODO
let (hostname_len, addr_len, tag_len, zone_len) = status
.iter()
.map(|(id, addr, _)| (addr, config.members.get(&adv.id)))
.map(|(addr, cfg)| {
(
8,
addr.to_string().len(),
cfg.map(|c| c.tag.len()).unwrap_or(0),
cfg.map(|c| c.zone.len()).unwrap_or(0),
)
})
.fold((0, 0, 0, 0), |(h, a, t, z), (mh, ma, mt, mz)| {
(max(h, mh), max(a, ma), max(t, mt), max(z, mz))
});
println!("Healthy nodes:"); println!("Healthy nodes:");
for (id, addr, _) in status.iter().filter(|(id, addr, is_up)| is_up) { let mut healthy_nodes = vec!["ID\tHostname\tAddress\tTag\tZone\tCapacity".to_string()];
for adv in status.iter().filter(|adv| adv.is_up) {
if let Some(cfg) = config.members.get(&adv.id) { if let Some(cfg) = config.members.get(&adv.id) {
println!( healthy_nodes.push(format!(
"{id:?}\t{host}{h_pad}\t{addr}{a_pad}\t[{tag}]{t_pad}\t{zone}{z_pad}\t{capacity}", "{id:?}\t{host}\t{addr}\t[{tag}]\t{zone}\t{capacity}",
id = id, id = adv.id,
host = "", host = adv.status.hostname,
addr = addr, addr = adv.addr,
tag = cfg.tag, tag = cfg.tag,
zone = cfg.zone, zone = cfg.zone,
capacity = cfg.capacity_string(), capacity = cfg.capacity_string(),
h_pad = " ".repeat(hostname_len - adv.state_info.hostname.len()), ));
a_pad = " ".repeat(addr_len - adv.addr.to_string().len()),
t_pad = " ".repeat(tag_len - cfg.tag.len()),
z_pad = " ".repeat(zone_len - cfg.zone.len()),
);
} else { } else {
println!( healthy_nodes.push(format!(
"{id:?}\t{h}{h_pad}\t{addr}{a_pad}\tUNCONFIGURED/REMOVED", "{id:?}\t{h}\t{addr}\tUNCONFIGURED/REMOVED",
id = id, id = adv.id,
h = "", h = adv.status.hostname,
addr = addr, addr = adv.addr,
h_pad = " ".repeat(hostname_len - "".len()), ));
a_pad = " ".repeat(addr_len - addr.to_string().len()),
);
} }
} }
format_table(healthy_nodes);
let status_keys = status.iter().map(|(id, _, _)| id).collect::<HashSet<_>>(); let status_keys = status.iter().map(|adv| adv.id).collect::<HashSet<_>>();
let failure_case_1 = status.iter().any(|(_, _, is_up)| !is_up); let failure_case_1 = status.iter().any(|adv| !adv.is_up);
let failure_case_2 = config let failure_case_2 = config
.members .members
.iter() .iter()
.any(|(id, _)| !status_keys.contains(id)); .any(|(id, _)| !status_keys.contains(id));
if failure_case_1 || failure_case_2 { if failure_case_1 || failure_case_2 {
println!("\nFailed nodes:"); println!("\nFailed nodes:");
for (id, addr) in status.iter().filter(|(_, _, is_up)| !is_up) { let mut failed_nodes = vec!["ID\tHostname\tAddress\tTag\tZone\tCapacity\tLast seen".to_string()];
if let Some(cfg) = config.members.get(&id) { for adv in status.iter().filter(|adv| !adv.is_up) {
println!( if let Some(cfg) = config.members.get(&adv.id) {
"{id:?}\t{host}{h_pad}\t{addr}{a_pad}\t[{tag}]{t_pad}\t{zone}{z_pad}\t{capacity}\tlast seen: {last_seen}s ago", failed_nodes.push(format!(
id=id, "{id:?}\t{host}\t{addr}\t[{tag}]\t{zone}\t{capacity}\t{last_seen}s ago",
host="", id = adv.id,
addr=addr, host = adv.status.hostname,
tag=cfg.tag, addr = adv.addr,
zone=cfg.zone, tag = cfg.tag,
capacity=cfg.capacity_string(), zone = cfg.zone,
last_seen=(now_msec() - 0) / 1000, capacity = cfg.capacity_string(),
h_pad=" ".repeat(hostname_len - "".len()), last_seen = (now_msec() - 0) / 1000,
a_pad=" ".repeat(addr_len - addr.to_string().len()), ));
t_pad=" ".repeat(tag_len - cfg.tag.len()),
z_pad=" ".repeat(zone_len - cfg.zone.len()),
);
} }
} }
let (tag_len, zone_len) = config
.members
.iter()
.filter(|(&id, _)| !status.iter().any(|(xid, _, _)| xid == id))
.map(|(_, cfg)| (cfg.tag.len(), cfg.zone.len()))
.fold((0, 0), |(t, z), (mt, mz)| (max(t, mt), max(z, mz)));
for (id, cfg) in config.members.iter() { for (id, cfg) in config.members.iter() {
if !status.iter().any(|(xid, _, _)| xid == *id) { if !status.iter().any(|adv| adv.id == *id) {
println!( failed_nodes.push(format!(
"{id:?}\t{tag}{t_pad}\t{zone}{z_pad}\t{capacity}\tnever seen", "{id:?}\t??\t??\t[{tag}]\t{zone}\t{capacity}\tnever seen",
id = id, id = id,
tag = cfg.tag, tag = cfg.tag,
zone = cfg.zone, zone = cfg.zone,
capacity = cfg.capacity_string(), capacity = cfg.capacity_string(),
t_pad = " ".repeat(tag_len - cfg.tag.len()), ));
z_pad = " ".repeat(zone_len - cfg.zone.len()),
);
} }
} }
format_table(failed_nodes);
} }
*/
Ok(()) Ok(())
} }
pub fn find_matching_node( pub async fn cmd_connect(
cand: impl std::iter::Iterator<Item = Uuid>, rpc_cli: &Endpoint<SystemRpc, ()>,
pattern: &str, rpc_host: NodeID,
) -> Result<Uuid, Error> { args: ConnectNodeOpt,
let mut candidates = vec![]; ) -> Result<(), Error> {
for c in cand { match rpc_cli.call(&rpc_host, &SystemRpc::Connect(args.node), PRIO_NORMAL).await?? {
if hex::encode(&c).starts_with(&pattern) { SystemRpc::Ok => {
candidates.push(c); println!("Success.");
Ok(())
} }
r => {
Err(Error::BadRpc(format!("Unexpected response: {:?}", r)))
} }
if candidates.len() != 1 {
Err(Error::Message(format!(
"{} nodes match '{}'",
candidates.len(),
pattern,
)))
} else {
Ok(candidates[0])
} }
} }
@ -472,22 +439,17 @@ pub async fn cmd_configure(
) -> Result<(), Error> { ) -> Result<(), Error> {
let status = match rpc_cli let status = match rpc_cli
.call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL) .call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL)
.await? .await??
{ {
SystemRpc::ReturnKnownNodes(nodes) => nodes, SystemRpc::ReturnKnownNodes(nodes) => nodes,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))), resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
}; };
let added_node = find_matching_node( let added_node = find_matching_node(status.iter().map(|adv| adv.id), &args.node_id)?;
status
.iter()
.map(|(id, _, _)| Uuid::try_from(id.as_ref()).unwrap()),
&args.node_id,
)?;
let mut config = match rpc_cli let mut config = match rpc_cli
.call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL) .call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL)
.await? .await??
{ {
SystemRpc::AdvertiseConfig(cfg) => cfg, SystemRpc::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))), resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
@ -544,7 +506,7 @@ pub async fn cmd_configure(
rpc_cli rpc_cli
.call(&rpc_host, &SystemRpc::AdvertiseConfig(config), PRIO_NORMAL) .call(&rpc_host, &SystemRpc::AdvertiseConfig(config), PRIO_NORMAL)
.await?; .await??;
Ok(()) Ok(())
} }
@ -555,7 +517,7 @@ pub async fn cmd_remove(
) -> Result<(), Error> { ) -> Result<(), Error> {
let mut config = match rpc_cli let mut config = match rpc_cli
.call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL) .call(&rpc_host, &SystemRpc::PullConfig, PRIO_NORMAL)
.await? .await??
{ {
SystemRpc::AdvertiseConfig(cfg) => cfg, SystemRpc::AdvertiseConfig(cfg) => cfg,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))), resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
@ -575,7 +537,7 @@ pub async fn cmd_remove(
rpc_cli rpc_cli
.call(&rpc_host, &SystemRpc::AdvertiseConfig(config), PRIO_NORMAL) .call(&rpc_host, &SystemRpc::AdvertiseConfig(config), PRIO_NORMAL)
.await?; .await??;
Ok(()) Ok(())
} }
@ -584,7 +546,7 @@ pub async fn cmd_admin(
rpc_host: NodeID, rpc_host: NodeID,
args: AdminRpc, args: AdminRpc,
) -> Result<(), Error> { ) -> Result<(), Error> {
match rpc_cli.call(&rpc_host, &args, PRIO_NORMAL).await? { match rpc_cli.call(&rpc_host, &args, PRIO_NORMAL).await?? {
AdminRpc::Ok(msg) => { AdminRpc::Ok(msg) => {
println!("{}", msg); println!("{}", msg);
} }
@ -613,6 +575,8 @@ pub async fn cmd_admin(
Ok(()) Ok(())
} }
// --- Utility functions ----
fn print_key_info(key: &Key) { fn print_key_info(key: &Key) {
println!("Key name: {}", key.name.get()); println!("Key name: {}", key.name.get());
println!("Key ID: {}", key.key_id); println!("Key ID: {}", key.key_id);
@ -640,3 +604,54 @@ fn print_bucket_info(bucket: &Bucket) {
} }
}; };
} }
fn format_table(data: Vec<String>) {
let data = data
.iter()
.map(|s| s.split('\t').collect::<Vec<_>>())
.collect::<Vec<_>>();
let columns = data.iter().map(|row| row.len()).fold(0, std::cmp::max);
let mut column_size = vec![0; columns];
let mut out = String::new();
for row in data.iter() {
for (i, col) in row.iter().enumerate() {
column_size[i] = std::cmp::max(column_size[i], col.chars().count());
}
}
for row in data.iter() {
for (col, col_len) in row[..row.len() - 1].iter().zip(column_size.iter()) {
out.push_str(col);
(0..col_len - col.chars().count() + 2).for_each(|_| out.push(' '));
}
out.push_str(&row[row.len() - 1]);
out.push('\n');
}
print!("{}", out);
}
pub fn find_matching_node(
cand: impl std::iter::Iterator<Item = Uuid>,
pattern: &str,
) -> Result<Uuid, Error> {
let mut candidates = vec![];
for c in cand {
if hex::encode(&c).starts_with(&pattern) {
candidates.push(c);
}
}
if candidates.len() != 1 {
Err(Error::Message(format!(
"{} nodes match '{}'",
candidates.len(),
pattern,
)))
} else {
Ok(candidates[0])
}
}

View file

@ -9,8 +9,6 @@ mod cli;
mod repair; mod repair;
mod server; mod server;
use std::net::SocketAddr;
use structopt::StructOpt; use structopt::StructOpt;
use netapp::util::parse_peer_addr; use netapp::util::parse_peer_addr;
@ -43,6 +41,7 @@ struct Opt {
#[tokio::main] #[tokio::main]
async fn main() { async fn main() {
pretty_env_logger::init(); pretty_env_logger::init();
sodiumoxide::init().expect("Unable to init sodiumoxide");
let opt = Opt::from_args(); let opt = Opt::from_args();

View file

@ -71,8 +71,14 @@ pub async fn run_server(config_file: PathBuf) -> Result<(), Error> {
// Remove RPC handlers for system to break reference cycles // Remove RPC handlers for system to break reference cycles
garage.system.netapp.drop_all_handlers(); garage.system.netapp.drop_all_handlers();
// Await for last parts to end // Await for netapp RPC system to end
run_system.await?; run_system.await?;
// Break last reference cycles so that stuff can terminate properly
garage.break_reference_cycles();
drop(garage);
// Await for all background tasks to end
await_background_done.await?; await_background_done.await?;
info!("Cleaning up..."); info!("Cleaning up...");

View file

@ -38,7 +38,6 @@ const RESYNC_RETRY_TIMEOUT: Duration = Duration::from_secs(10);
#[derive(Debug, Serialize, Deserialize)] #[derive(Debug, Serialize, Deserialize)]
pub enum BlockRpc { pub enum BlockRpc {
Ok, Ok,
Error(String),
/// Message to ask for a block of data, by hash /// Message to ask for a block of data, by hash
GetBlock(Hash), GetBlock(Hash),
/// Message to send a block of data, either because requested, of for first delivery of new /// Message to send a block of data, either because requested, of for first delivery of new
@ -61,8 +60,8 @@ pub struct PutBlockMessage {
pub data: Vec<u8>, pub data: Vec<u8>,
} }
impl Message for BlockRpc { impl Rpc for BlockRpc {
type Response = BlockRpc; type Response = Result<BlockRpc, Error>;
} }
/// The block manager, handling block exchange between nodes, and block storage on local node /// The block manager, handling block exchange between nodes, and block storage on local node
@ -117,15 +116,6 @@ impl BlockManager {
block_manager block_manager
} }
async fn handle_rpc(self: Arc<Self>, msg: &BlockRpc) -> Result<BlockRpc, Error> {
match msg {
BlockRpc::PutBlock(m) => self.write_block(&m.hash, &m.data).await,
BlockRpc::GetBlock(h) => self.read_block(h).await,
BlockRpc::NeedBlockQuery(h) => self.need_block(h).await.map(BlockRpc::NeedBlockReply),
_ => Err(Error::BadRpc("Unexpected RPC message".to_string())),
}
}
pub fn spawn_background_worker(self: Arc<Self>) { pub fn spawn_background_worker(self: Arc<Self>) {
// Launch 2 simultaneous workers for background resync loop preprocessing <= TODO actually this // Launch 2 simultaneous workers for background resync loop preprocessing <= TODO actually this
// launches only one worker with current value of BACKGROUND_WORKERS // launches only one worker with current value of BACKGROUND_WORKERS
@ -532,11 +522,17 @@ impl BlockManager {
#[async_trait] #[async_trait]
impl EndpointHandler<BlockRpc> for BlockManager { impl EndpointHandler<BlockRpc> for BlockManager {
async fn handle(self: &Arc<Self>, message: &BlockRpc, _from: NodeID) -> BlockRpc { async fn handle(
self.clone() self: &Arc<Self>,
.handle_rpc(message) message: &BlockRpc,
.await _from: NodeID,
.unwrap_or_else(|e| BlockRpc::Error(format!("{}", e))) ) -> Result<BlockRpc, Error> {
match message {
BlockRpc::PutBlock(m) => self.write_block(&m.hash, &m.data).await,
BlockRpc::GetBlock(h) => self.read_block(h).await,
BlockRpc::NeedBlockQuery(h) => self.need_block(h).await.map(BlockRpc::NeedBlockReply),
_ => Err(Error::BadRpc("Unexpected RPC message".to_string())),
}
} }
} }

View file

@ -61,6 +61,7 @@ impl Garage {
background.clone(), background.clone(),
replication_mode.replication_factor(), replication_mode.replication_factor(),
config.rpc_bind_addr, config.rpc_bind_addr,
config.rpc_public_addr,
config.bootstrap_peers.clone(), config.bootstrap_peers.clone(),
config.consul_host.clone(), config.consul_host.clone(),
config.consul_service_name.clone(), config.consul_service_name.clone(),
@ -162,4 +163,9 @@ impl Garage {
garage garage
} }
/// Use this for shutdown
pub fn break_reference_cycles(&self) {
self.block_manager.garage.swap(None);
}
} }

View file

@ -15,8 +15,6 @@ path = "lib.rs"
[dependencies] [dependencies]
garage_util = { version = "0.4.0", path = "../util" } garage_util = { version = "0.4.0", path = "../util" }
garage_rpc_021 = { package = "garage_rpc", version = "0.2.1" }
arc-swap = "1.0" arc-swap = "1.0"
bytes = "1.0" bytes = "1.0"
gethostname = "0.2" gethostname = "0.2"
@ -36,5 +34,5 @@ tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi
tokio-stream = { version = "0.1", features = ["net"] } tokio-stream = { version = "0.1", features = ["net"] }
netapp = { version = "0.3.0", git = "https://git.deuxfleurs.fr/lx/netapp" } netapp = { version = "0.3.0", git = "https://git.deuxfleurs.fr/lx/netapp" }
hyper = "0.14" hyper = { version = "0.14", features = ["client", "http1", "runtime", "tcp"] }

View file

@ -1,24 +1,31 @@
use std::collections::HashMap;
use std::net::{IpAddr, SocketAddr}; use std::net::{IpAddr, SocketAddr};
use hyper::client::Client; use hyper::client::Client;
use hyper::StatusCode; use hyper::StatusCode;
use hyper::{Body, Method, Request}; use hyper::{Body, Method, Request};
use serde::Deserialize; use serde::{Deserialize, Serialize};
use netapp::NodeID;
use garage_util::error::Error; use garage_util::error::Error;
#[derive(Deserialize, Clone)] // ---- READING FROM CONSUL CATALOG ----
struct ConsulEntry {
#[serde(alias = "Address")] #[derive(Deserialize, Clone, Debug)]
struct ConsulQueryEntry {
#[serde(rename = "Address")]
address: String, address: String,
#[serde(alias = "ServicePort")] #[serde(rename = "ServicePort")]
service_port: u16, service_port: u16,
#[serde(rename = "NodeMeta")]
node_meta: HashMap<String, String>,
} }
pub async fn get_consul_nodes( pub async fn get_consul_nodes(
consul_host: &str, consul_host: &str,
consul_service_name: &str, consul_service_name: &str,
) -> Result<Vec<SocketAddr>, Error> { ) -> Result<Vec<(NodeID, SocketAddr)>, Error> {
let url = format!( let url = format!(
"http://{}/v1/catalog/service/{}", "http://{}/v1/catalog/service/{}",
consul_host, consul_service_name consul_host, consul_service_name
@ -36,17 +43,111 @@ pub async fn get_consul_nodes(
} }
let body = hyper::body::to_bytes(resp.into_body()).await?; let body = hyper::body::to_bytes(resp.into_body()).await?;
let entries = serde_json::from_slice::<Vec<ConsulEntry>>(body.as_ref())?; let entries = serde_json::from_slice::<Vec<ConsulQueryEntry>>(body.as_ref())?;
let mut ret = vec![]; let mut ret = vec![];
for ent in entries { for ent in entries {
let ip = ent let ip = ent.address.parse::<IpAddr>().ok();
.address let pubkey = ent
.parse::<IpAddr>() .node_meta
.map_err(|e| Error::Message(format!("Could not parse IP address: {}", e)))?; .get("pubkey")
ret.push(SocketAddr::new(ip, ent.service_port)); .map(|k| hex::decode(&k).ok())
.flatten()
.map(|k| NodeID::from_slice(&k[..]))
.flatten();
if let (Some(ip), Some(pubkey)) = (ip, pubkey) {
ret.push((pubkey, SocketAddr::new(ip, ent.service_port)));
} else {
warn!(
"Could not process node spec from Consul: {:?} (invalid IP or public key)",
ent
);
}
} }
debug!("Got nodes from Consul: {:?}", ret); debug!("Got nodes from Consul: {:?}", ret);
Ok(ret) Ok(ret)
} }
// ---- PUBLISHING TO CONSUL CATALOG ----
#[derive(Serialize, Clone, Debug)]
struct ConsulPublishEntry {
#[serde(rename = "Node")]
node: String,
#[serde(rename = "Address")]
address: IpAddr,
#[serde(rename = "NodeMeta")]
node_meta: HashMap<String, String>,
#[serde(rename = "Service")]
service: ConsulPublishService,
}
#[derive(Serialize, Clone, Debug)]
struct ConsulPublishService {
#[serde(rename = "ID")]
service_id: String,
#[serde(rename = "Service")]
service_name: String,
#[serde(rename = "Tags")]
tags: Vec<String>,
#[serde(rename = "Address")]
address: IpAddr,
#[serde(rename = "Port")]
port: u16,
}
pub async fn publish_consul_service(
consul_host: &str,
consul_service_name: &str,
node_id: NodeID,
hostname: &str,
rpc_public_addr: SocketAddr,
) -> Result<(), Error> {
let node = format!("garage:{}", hex::encode(&node_id[..8]));
let advertisment = ConsulPublishEntry {
node: node.clone(),
address: rpc_public_addr.ip(),
node_meta: [
("pubkey".to_string(), hex::encode(node_id)),
("hostname".to_string(), hostname.to_string()),
]
.iter()
.cloned()
.collect(),
service: ConsulPublishService {
service_id: node.clone(),
service_name: consul_service_name.to_string(),
tags: vec!["advertised-by-garage".into(), hostname.into()],
address: rpc_public_addr.ip(),
port: rpc_public_addr.port(),
},
};
let url = format!("http://{}/v1/catalog/register", consul_host);
let req_body = serde_json::to_string(&advertisment)?;
debug!("Request body for consul adv: {}", req_body);
let req = Request::builder()
.uri(url)
.method(Method::PUT)
.body(Body::from(req_body))?;
let client = Client::new();
let resp = client.request(req).await?;
debug!("Response of advertising to Consul: {:?}", resp);
let resp_code = resp.status();
debug!(
"{}",
std::str::from_utf8(&hyper::body::to_bytes(resp.into_body()).await?)
.unwrap_or("<invalid utf8>")
);
if resp_code != StatusCode::OK {
return Err(Error::Message(format!("HTTP error {}", resp_code)));
}
Ok(())
}

View file

@ -3,8 +3,6 @@
use std::collections::{HashMap, HashSet}; use std::collections::{HashMap, HashSet};
use std::convert::TryInto; use std::convert::TryInto;
use netapp::NodeID;
use serde::{Deserialize, Serialize}; use serde::{Deserialize, Serialize};
use garage_util::data::*; use garage_util::data::*;
@ -40,31 +38,6 @@ impl NetworkConfig {
version: 0, version: 0,
} }
} }
pub(crate) fn migrate_from_021(old: garage_rpc_021::ring::NetworkConfig) -> Self {
let members = old
.members
.into_iter()
.map(|(id, conf)| {
(
Hash::try_from(id.as_slice()).unwrap(),
NetworkConfigEntry {
zone: conf.datacenter,
capacity: if conf.capacity == 0 {
None
} else {
Some(conf.capacity)
},
tag: conf.tag,
},
)
})
.collect();
Self {
members,
version: old.version,
}
}
} }
/// The overall configuration of one (possibly remote) node /// The overall configuration of one (possibly remote) node
@ -100,7 +73,7 @@ pub struct Ring {
pub config: NetworkConfig, pub config: NetworkConfig,
// Internal order of nodes used to make a more compact representation of the ring // Internal order of nodes used to make a more compact representation of the ring
nodes: Vec<NodeID>, nodes: Vec<Uuid>,
// The list of entries in the ring // The list of entries in the ring
ring: Vec<RingEntry>, ring: Vec<RingEntry>,
@ -262,11 +235,6 @@ impl Ring {
}) })
.collect::<Vec<_>>(); .collect::<Vec<_>>();
let nodes = nodes
.iter()
.map(|id| NodeID::from_slice(id.as_slice()).unwrap())
.collect::<Vec<_>>();
Self { Self {
replication_factor, replication_factor,
config, config,
@ -298,7 +266,7 @@ impl Ring {
} }
/// Walk the ring to find the n servers in which data should be replicated /// Walk the ring to find the n servers in which data should be replicated
pub fn get_nodes(&self, position: &Hash, n: usize) -> Vec<NodeID> { pub fn get_nodes(&self, position: &Hash, n: usize) -> Vec<Uuid> {
if self.ring.len() != 1 << PARTITION_BITS { if self.ring.len() != 1 << PARTITION_BITS {
warn!("Ring not yet ready, read/writes will be lost!"); warn!("Ring not yet ready, read/writes will be lost!");
return vec![]; return vec![];

View file

@ -8,13 +8,14 @@ use futures::stream::StreamExt;
use futures_util::future::FutureExt; use futures_util::future::FutureExt;
use tokio::select; use tokio::select;
pub use netapp::endpoint::{Endpoint, EndpointHandler, Message}; pub use netapp::endpoint::{Endpoint, EndpointHandler, Message as Rpc};
use netapp::peering::fullmesh::FullMeshPeeringStrategy; use netapp::peering::fullmesh::FullMeshPeeringStrategy;
pub use netapp::proto::*; pub use netapp::proto::*;
pub use netapp::{NetApp, NodeID}; pub use netapp::{NetApp, NodeID};
use garage_util::background::BackgroundRunner; use garage_util::background::BackgroundRunner;
use garage_util::error::{Error, RpcError}; use garage_util::error::Error;
use garage_util::data::Uuid;
const DEFAULT_TIMEOUT: Duration = Duration::from_secs(10); const DEFAULT_TIMEOUT: Duration = Duration::from_secs(10);
@ -66,46 +67,47 @@ pub struct RpcHelper {
} }
impl RpcHelper { impl RpcHelper {
pub async fn call<M, H>( pub async fn call<M, H, S>(
&self, &self,
endpoint: &Endpoint<M, H>, endpoint: &Endpoint<M, H>,
to: NodeID, to: Uuid,
msg: M, msg: M,
strat: RequestStrategy, strat: RequestStrategy,
) -> Result<M::Response, Error> ) -> Result<S, Error>
where where
M: Message, M: Rpc<Response = Result<S, Error>>,
H: EndpointHandler<M>, H: EndpointHandler<M>,
{ {
self.call_arc(endpoint, to, Arc::new(msg), strat).await self.call_arc(endpoint, to, Arc::new(msg), strat).await
} }
pub async fn call_arc<M, H>( pub async fn call_arc<M, H, S>(
&self, &self,
endpoint: &Endpoint<M, H>, endpoint: &Endpoint<M, H>,
to: NodeID, to: Uuid,
msg: Arc<M>, msg: Arc<M>,
strat: RequestStrategy, strat: RequestStrategy,
) -> Result<M::Response, Error> ) -> Result<S, Error>
where where
M: Message, M: Rpc<Response = Result<S, Error>>,
H: EndpointHandler<M>, H: EndpointHandler<M>,
{ {
let node_id = to.into();
select! { select! {
res = endpoint.call(&to, &msg, strat.rs_priority) => Ok(res?), res = endpoint.call(&node_id, &msg, strat.rs_priority) => Ok(res??),
_ = tokio::time::sleep(strat.rs_timeout) => Err(Error::Rpc(RpcError::Timeout)), _ = tokio::time::sleep(strat.rs_timeout) => Err(Error::Timeout),
} }
} }
pub async fn call_many<M, H>( pub async fn call_many<M, H, S>(
&self, &self,
endpoint: &Endpoint<M, H>, endpoint: &Endpoint<M, H>,
to: &[NodeID], to: &[Uuid],
msg: M, msg: M,
strat: RequestStrategy, strat: RequestStrategy,
) -> Vec<(NodeID, Result<M::Response, Error>)> ) -> Vec<(Uuid, Result<S, Error>)>
where where
M: Message, M: Rpc<Response = Result<S, Error>>,
H: EndpointHandler<M>, H: EndpointHandler<M>,
{ {
let msg = Arc::new(msg); let msg = Arc::new(msg);
@ -120,37 +122,38 @@ impl RpcHelper {
.collect::<Vec<_>>() .collect::<Vec<_>>()
} }
pub async fn broadcast<M, H>( pub async fn broadcast<M, H, S>(
&self, &self,
endpoint: &Endpoint<M, H>, endpoint: &Endpoint<M, H>,
msg: M, msg: M,
strat: RequestStrategy, strat: RequestStrategy,
) -> Vec<(NodeID, Result<M::Response, Error>)> ) -> Vec<(Uuid, Result<S, Error>)>
where where
M: Message, M: Rpc<Response = Result<S, Error>>,
H: EndpointHandler<M>, H: EndpointHandler<M>,
{ {
let to = self let to = self
.fullmesh .fullmesh
.get_peer_list() .get_peer_list()
.iter() .iter()
.map(|p| p.id) .map(|p| p.id.into())
.collect::<Vec<_>>(); .collect::<Vec<_>>();
self.call_many(endpoint, &to[..], msg, strat).await self.call_many(endpoint, &to[..], msg, strat).await
} }
/// Make a RPC call to multiple servers, returning either a Vec of responses, or an error if /// Make a RPC call to multiple servers, returning either a Vec of responses, or an error if
/// strategy could not be respected due to too many errors /// strategy could not be respected due to too many errors
pub async fn try_call_many<M, H>( pub async fn try_call_many<M, H, S>(
&self, &self,
endpoint: &Arc<Endpoint<M, H>>, endpoint: &Arc<Endpoint<M, H>>,
to: &[NodeID], to: &[Uuid],
msg: M, msg: M,
strategy: RequestStrategy, strategy: RequestStrategy,
) -> Result<Vec<M::Response>, Error> ) -> Result<Vec<S>, Error>
where where
M: Message + 'static, M: Rpc<Response = Result<S, Error>> + 'static,
H: EndpointHandler<M> + 'static, H: EndpointHandler<M> + 'static,
S: Send,
{ {
let msg = Arc::new(msg); let msg = Arc::new(msg);
let mut resp_stream = to let mut resp_stream = to
@ -200,7 +203,7 @@ impl RpcHelper {
Ok(results) Ok(results)
} else { } else {
let errors = errors.iter().map(|e| format!("{}", e)).collect::<Vec<_>>(); let errors = errors.iter().map(|e| format!("{}", e)).collect::<Vec<_>>();
Err(Error::from(RpcError::TooManyErrors(errors))) Err(Error::TooManyErrors(errors))
} }
} }
} }

View file

@ -1,8 +1,9 @@
//! Module containing structs related to membership management //! Module containing structs related to membership management
use std::collections::HashMap;
use std::io::{Read, Write}; use std::io::{Read, Write};
use std::net::SocketAddr; use std::net::SocketAddr;
use std::path::{Path, PathBuf}; use std::path::{Path, PathBuf};
use std::sync::Arc; use std::sync::{Arc, RwLock};
use std::time::Duration; use std::time::Duration;
use arc_swap::ArcSwap; use arc_swap::ArcSwap;
@ -14,21 +15,24 @@ use sodiumoxide::crypto::sign::ed25519;
use tokio::sync::watch; use tokio::sync::watch;
use tokio::sync::Mutex; use tokio::sync::Mutex;
use netapp::endpoint::{Endpoint, EndpointHandler, Message}; use netapp::endpoint::{Endpoint, EndpointHandler};
use netapp::peering::fullmesh::FullMeshPeeringStrategy; use netapp::peering::fullmesh::FullMeshPeeringStrategy;
use netapp::proto::*; use netapp::proto::*;
use netapp::{NetApp, NetworkKey, NodeID, NodeKey}; use netapp::{NetApp, NetworkKey, NodeID, NodeKey};
use netapp::util::parse_and_resolve_peer_addr;
use garage_util::background::BackgroundRunner; use garage_util::background::BackgroundRunner;
use garage_util::data::Uuid;
use garage_util::error::Error; use garage_util::error::Error;
use garage_util::persister::Persister; use garage_util::persister::Persister;
//use garage_util::time::*; use garage_util::time::*;
//use crate::consul::get_consul_nodes; use crate::consul::*;
use crate::ring::*; use crate::ring::*;
use crate::rpc_helper::{RequestStrategy, RpcHelper}; use crate::rpc_helper::*;
const DISCOVERY_INTERVAL: Duration = Duration::from_secs(60); const DISCOVERY_INTERVAL: Duration = Duration::from_secs(60);
const STATUS_EXCHANGE_INTERVAL: Duration = Duration::from_secs(10);
const PING_TIMEOUT: Duration = Duration::from_secs(2); const PING_TIMEOUT: Duration = Duration::from_secs(2);
/// RPC endpoint used for calls related to membership /// RPC endpoint used for calls related to membership
@ -39,33 +43,35 @@ pub const SYSTEM_RPC_PATH: &str = "garage_rpc/membership.rs/SystemRpc";
pub enum SystemRpc { pub enum SystemRpc {
/// Response to successfull advertisements /// Response to successfull advertisements
Ok, Ok,
/// Error response /// Request to connect to a specific node (in <pubkey>@<host>:<port> format)
Error(String), Connect(String),
/// Ask other node its config. Answered with AdvertiseConfig /// Ask other node its config. Answered with AdvertiseConfig
PullConfig, PullConfig,
/// Advertise Garage status. Answered with another AdvertiseStatus. /// Advertise Garage status. Answered with another AdvertiseStatus.
/// Exchanged with every node on a regular basis. /// Exchanged with every node on a regular basis.
AdvertiseStatus(StateInfo), AdvertiseStatus(NodeStatus),
/// Advertisement of nodes config. Sent spontanously or in response to PullConfig /// Advertisement of nodes config. Sent spontanously or in response to PullConfig
AdvertiseConfig(NetworkConfig), AdvertiseConfig(NetworkConfig),
/// Get known nodes states /// Get known nodes states
GetKnownNodes, GetKnownNodes,
/// Return known nodes /// Return known nodes
ReturnKnownNodes(Vec<(NodeID, SocketAddr, bool)>), ReturnKnownNodes(Vec<KnownNodeInfo>),
} }
impl Message for SystemRpc { impl Rpc for SystemRpc {
type Response = SystemRpc; type Response = Result<SystemRpc, Error>;
} }
/// This node's membership manager /// This node's membership manager
pub struct System { pub struct System {
/// The id of this node /// The id of this node
pub id: NodeID, pub id: Uuid,
persist_config: Persister<NetworkConfig>, persist_config: Persister<NetworkConfig>,
persist_peer_list: Persister<Vec<(Uuid, SocketAddr)>>,
state_info: ArcSwap<StateInfo>, local_status: ArcSwap<NodeStatus>,
node_status: RwLock<HashMap<Uuid, (u64, NodeStatus)>>,
pub netapp: Arc<NetApp>, pub netapp: Arc<NetApp>,
fullmesh: Arc<FullMeshPeeringStrategy>, fullmesh: Arc<FullMeshPeeringStrategy>,
@ -74,6 +80,7 @@ pub struct System {
system_endpoint: Arc<Endpoint<SystemRpc, System>>, system_endpoint: Arc<Endpoint<SystemRpc, System>>,
rpc_listen_addr: SocketAddr, rpc_listen_addr: SocketAddr,
rpc_public_addr: Option<SocketAddr>,
bootstrap_peers: Vec<(NodeID, SocketAddr)>, bootstrap_peers: Vec<(NodeID, SocketAddr)>,
consul_host: Option<String>, consul_host: Option<String>,
consul_service_name: Option<String>, consul_service_name: Option<String>,
@ -88,7 +95,7 @@ pub struct System {
} }
#[derive(Debug, Clone, Serialize, Deserialize)] #[derive(Debug, Clone, Serialize, Deserialize)]
pub struct StateInfo { pub struct NodeStatus {
/// Hostname of the node /// Hostname of the node
pub hostname: String, pub hostname: String,
/// Replication factor configured on the node /// Replication factor configured on the node
@ -97,26 +104,34 @@ pub struct StateInfo {
pub config_version: u64, pub config_version: u64,
} }
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct KnownNodeInfo {
pub id: Uuid,
pub addr: SocketAddr,
pub is_up: bool,
pub status: NodeStatus,
}
fn gen_node_key(metadata_dir: &Path) -> Result<NodeKey, Error> { fn gen_node_key(metadata_dir: &Path) -> Result<NodeKey, Error> {
let mut id_file = metadata_dir.to_path_buf(); let mut key_file = metadata_dir.to_path_buf();
id_file.push("node_id"); key_file.push("node_key");
if id_file.as_path().exists() { if key_file.as_path().exists() {
let mut f = std::fs::File::open(id_file.as_path())?; let mut f = std::fs::File::open(key_file.as_path())?;
let mut d = vec![]; let mut d = vec![];
f.read_to_end(&mut d)?; f.read_to_end(&mut d)?;
if d.len() != 64 { if d.len() != 64 {
return Err(Error::Message("Corrupt node_id file".to_string())); return Err(Error::Message("Corrupt node_key file".to_string()));
} }
let mut key = [0u8; 64]; let mut key = [0u8; 64];
key.copy_from_slice(&d[..]); key.copy_from_slice(&d[..]);
Ok(NodeKey::from_slice(&key[..]).unwrap()) Ok(NodeKey::from_slice(&key[..]).unwrap())
} else { } else {
let (key, _) = ed25519::gen_keypair(); let (_, key) = ed25519::gen_keypair();
let mut f = std::fs::File::create(id_file.as_path())?; let mut f = std::fs::File::create(key_file.as_path())?;
f.write_all(&key[..])?; f.write_all(&key[..])?;
Ok(NodeKey::from_slice(&key[..]).unwrap()) Ok(key)
} }
} }
@ -128,6 +143,7 @@ impl System {
background: Arc<BackgroundRunner>, background: Arc<BackgroundRunner>,
replication_factor: usize, replication_factor: usize,
rpc_listen_addr: SocketAddr, rpc_listen_addr: SocketAddr,
rpc_public_address: Option<SocketAddr>,
bootstrap_peers: Vec<(NodeID, SocketAddr)>, bootstrap_peers: Vec<(NodeID, SocketAddr)>,
consul_host: Option<String>, consul_host: Option<String>,
consul_service_name: Option<String>, consul_service_name: Option<String>,
@ -136,29 +152,20 @@ impl System {
info!("Node public key: {}", hex::encode(&node_key.public_key())); info!("Node public key: {}", hex::encode(&node_key.public_key()));
let persist_config = Persister::new(&metadata_dir, "network_config"); let persist_config = Persister::new(&metadata_dir, "network_config");
let persist_peer_list = Persister::new(&metadata_dir, "peer_list");
let net_config = match persist_config.load() { let net_config = match persist_config.load() {
Ok(x) => x, Ok(x) => x,
Err(e) => { Err(e) => {
match Persister::<garage_rpc_021::ring::NetworkConfig>::new(
&metadata_dir,
"network_config",
)
.load()
{
Ok(old_config) => NetworkConfig::migrate_from_021(old_config),
Err(e2) => {
info!( info!(
"No valid previous network configuration stored ({}, {}), starting fresh.", "No valid previous network configuration stored ({}), starting fresh.",
e, e2 e
); );
NetworkConfig::new() NetworkConfig::new()
} }
}
}
}; };
let state_info = StateInfo { let local_status = NodeStatus {
hostname: gethostname::gethostname() hostname: gethostname::gethostname()
.into_string() .into_string()
.unwrap_or_else(|_| "<invalid utf-8>".to_string()), .unwrap_or_else(|_| "<invalid utf-8>".to_string()),
@ -169,15 +176,27 @@ impl System {
let ring = Ring::new(net_config, replication_factor); let ring = Ring::new(net_config, replication_factor);
let (update_ring, ring) = watch::channel(Arc::new(ring)); let (update_ring, ring) = watch::channel(Arc::new(ring));
if let Some(addr) = rpc_public_address {
println!("{}@{}", hex::encode(&node_key.public_key()), addr);
} else {
println!("{}", hex::encode(&node_key.public_key()));
}
let netapp = NetApp::new(network_key, node_key); let netapp = NetApp::new(network_key, node_key);
let fullmesh = FullMeshPeeringStrategy::new(netapp.clone(), bootstrap_peers.clone()); let fullmesh = FullMeshPeeringStrategy::new(
netapp.clone(),
bootstrap_peers.clone(),
rpc_public_address,
);
let system_endpoint = netapp.endpoint(SYSTEM_RPC_PATH.into()); let system_endpoint = netapp.endpoint(SYSTEM_RPC_PATH.into());
let sys = Arc::new(System { let sys = Arc::new(System {
id: netapp.id.clone(), id: netapp.id.into(),
persist_config, persist_config,
state_info: ArcSwap::new(Arc::new(state_info)), persist_peer_list,
local_status: ArcSwap::new(Arc::new(local_status)),
node_status: RwLock::new(HashMap::new()),
netapp: netapp.clone(), netapp: netapp.clone(),
fullmesh: fullmesh.clone(), fullmesh: fullmesh.clone(),
rpc: RpcHelper { rpc: RpcHelper {
@ -187,6 +206,7 @@ impl System {
system_endpoint, system_endpoint,
replication_factor, replication_factor,
rpc_listen_addr, rpc_listen_addr,
rpc_public_addr: rpc_public_address,
bootstrap_peers, bootstrap_peers,
consul_host, consul_host,
consul_service_name, consul_service_name,
@ -206,11 +226,38 @@ impl System {
.listen(self.rpc_listen_addr, None, must_exit.clone()), .listen(self.rpc_listen_addr, None, must_exit.clone()),
self.fullmesh.clone().run(must_exit.clone()), self.fullmesh.clone().run(must_exit.clone()),
self.discovery_loop(must_exit.clone()), self.discovery_loop(must_exit.clone()),
self.status_exchange_loop(must_exit.clone()),
); );
} }
// ---- INTERNALS ---- // ---- INTERNALS ----
async fn advertise_to_consul(self: Arc<Self>) -> Result<(), Error> {
let (consul_host, consul_service_name) =
match (&self.consul_host, &self.consul_service_name) {
(Some(ch), Some(csn)) => (ch, csn),
_ => return Ok(()),
};
let rpc_public_addr = match self.rpc_public_addr {
Some(addr) => addr,
None => {
warn!("Not advertising to Consul because rpc_public_addr is not defined in config file.");
return Ok(());
}
};
publish_consul_service(
consul_host,
consul_service_name,
self.netapp.id,
&self.local_status.load_full().hostname,
rpc_public_addr,
)
.await
.map_err(|e| Error::Message(format!("Error while publishing Consul service: {}", e)))
}
/// Save network configuration to disc /// Save network configuration to disc
async fn save_network_config(self: Arc<Self>) -> Result<(), Error> { async fn save_network_config(self: Arc<Self>) -> Result<(), Error> {
let ring: Arc<Ring> = self.ring.borrow().clone(); let ring: Arc<Ring> = self.ring.borrow().clone();
@ -221,12 +268,27 @@ impl System {
Ok(()) Ok(())
} }
fn update_state_info(&self) { fn update_local_status(&self) {
let mut new_si: StateInfo = self.state_info.load().as_ref().clone(); let mut new_si: NodeStatus = self.local_status.load().as_ref().clone();
let ring = self.ring.borrow(); let ring = self.ring.borrow();
new_si.config_version = ring.config.version; new_si.config_version = ring.config.version;
self.state_info.swap(Arc::new(new_si)); self.local_status.swap(Arc::new(new_si));
}
async fn handle_connect(&self, node: &str) -> Result<SystemRpc, Error> {
let (pubkey, addrs) = parse_and_resolve_peer_addr(node)
.ok_or_else(|| Error::Message(format!("Unable to parse or resolve node specification: {}", node)))?;
let mut errors = vec![];
for ip in addrs.iter() {
match self.netapp.clone().try_connect(*ip, pubkey).await {
Ok(()) => return Ok(SystemRpc::Ok),
Err(e) => {
errors.push((*ip, e));
}
}
}
return Err(Error::Message(format!("Could not connect to specified peers. Errors: {:?}", errors)));
} }
fn handle_pull_config(&self) -> SystemRpc { fn handle_pull_config(&self) -> SystemRpc {
@ -234,6 +296,58 @@ impl System {
SystemRpc::AdvertiseConfig(ring.config.clone()) SystemRpc::AdvertiseConfig(ring.config.clone())
} }
fn handle_get_known_nodes(&self) -> SystemRpc {
let node_status = self.node_status.read().unwrap();
let known_nodes =
self.fullmesh
.get_peer_list()
.iter()
.map(|n| KnownNodeInfo {
id: n.id.into(),
addr: n.addr,
is_up: n.is_up(),
status: node_status.get(&n.id.into()).cloned().map(|(_, st)| st).unwrap_or(
NodeStatus {
hostname: "?".to_string(),
replication_factor: 0,
config_version: 0,
},
),
})
.collect::<Vec<_>>();
SystemRpc::ReturnKnownNodes(known_nodes)
}
async fn handle_advertise_status(
self: &Arc<Self>,
from: Uuid,
info: &NodeStatus,
) -> Result<SystemRpc, Error> {
let local_info = self.local_status.load();
if local_info.replication_factor < info.replication_factor {
error!("Some node have a higher replication factor ({}) than this one ({}). This is not supported and might lead to bugs",
info.replication_factor,
local_info.replication_factor);
std::process::exit(1);
}
if info.config_version > local_info.config_version {
let self2 = self.clone();
self.background.spawn_cancellable(async move {
self2.pull_config(from).await;
Ok(())
});
}
self.node_status
.write()
.unwrap()
.insert(from, (now_msec(), info.clone()));
Ok(SystemRpc::Ok)
}
async fn handle_advertise_config( async fn handle_advertise_config(
self: Arc<Self>, self: Arc<Self>,
adv: &NetworkConfig, adv: &NetworkConfig,
@ -265,13 +379,32 @@ impl System {
Ok(SystemRpc::Ok) Ok(SystemRpc::Ok)
} }
async fn discovery_loop(&self, mut stop_signal: watch::Receiver<bool>) { async fn status_exchange_loop(&self, mut stop_signal: watch::Receiver<bool>) {
/* TODO while !*stop_signal.borrow() {
let restart_at = tokio::time::sleep(STATUS_EXCHANGE_INTERVAL);
self.update_local_status();
let local_status: NodeStatus = self.local_status.load().as_ref().clone();
self.rpc
.broadcast(
&self.system_endpoint,
SystemRpc::AdvertiseStatus(local_status),
RequestStrategy::with_priority(PRIO_HIGH).with_timeout(PING_TIMEOUT),
)
.await;
select! {
_ = restart_at.fuse() => {},
_ = stop_signal.changed().fuse() => {},
}
}
}
async fn discovery_loop(self: &Arc<Self>, mut stop_signal: watch::Receiver<bool>) {
let consul_config = match (&self.consul_host, &self.consul_service_name) { let consul_config = match (&self.consul_host, &self.consul_service_name) {
(Some(ch), Some(csn)) => Some((ch.clone(), csn.clone())), (Some(ch), Some(csn)) => Some((ch.clone(), csn.clone())),
_ => None, _ => None,
}; };
*/
while !*stop_signal.borrow() { while !*stop_signal.borrow() {
let not_configured = self.ring.borrow().config.members.is_empty(); let not_configured = self.ring.borrow().config.members.is_empty();
@ -286,34 +419,42 @@ impl System {
if not_configured || no_peers || bad_peers { if not_configured || no_peers || bad_peers {
info!("Doing a bootstrap/discovery step (not_configured: {}, no_peers: {}, bad_peers: {})", not_configured, no_peers, bad_peers); info!("Doing a bootstrap/discovery step (not_configured: {}, no_peers: {}, bad_peers: {})", not_configured, no_peers, bad_peers);
let ping_list = self.bootstrap_peers.clone(); let mut ping_list = self.bootstrap_peers.clone();
/* // Add peer list from list stored on disk
*TODO bring this back: persisted list of peers if let Ok(peers) = self.persist_peer_list.load_async().await {
if let Ok(peers) = self.persist_status.load_async().await { ping_list.extend(peers.iter().map(|(id, addr)| ((*id).into(), *addr)))
ping_list.extend(peers.iter().map(|x| (x.addr, Some(x.id))));
} }
*/
/* // Fetch peer list from Consul
* TODO bring this back: get peers from consul
if let Some((consul_host, consul_service_name)) = &consul_config { if let Some((consul_host, consul_service_name)) = &consul_config {
match get_consul_nodes(consul_host, consul_service_name).await { match get_consul_nodes(consul_host, consul_service_name).await {
Ok(node_list) => { Ok(node_list) => {
ping_list.extend(node_list.iter().map(|a| (*a, None))); ping_list.extend(node_list);
} }
Err(e) => { Err(e) => {
warn!("Could not retrieve node list from Consul: {}", e); warn!("Could not retrieve node list from Consul: {}", e);
} }
} }
} }
*/
for (node_id, node_addr) in ping_list { for (node_id, node_addr) in ping_list {
tokio::spawn(self.netapp.clone().try_connect(node_addr, node_id)); tokio::spawn(self.netapp.clone().try_connect(node_addr, node_id));
} }
} }
let peer_list = self
.fullmesh
.get_peer_list()
.iter()
.map(|n| (n.id.into(), n.addr))
.collect::<Vec<_>>();
if let Err(e) = self.persist_peer_list.save_async(&peer_list).await {
warn!("Could not save peer list to file: {}", e);
}
self.background.spawn(self.clone().advertise_to_consul());
let restart_at = tokio::time::sleep(DISCOVERY_INTERVAL); let restart_at = tokio::time::sleep(DISCOVERY_INTERVAL);
select! { select! {
_ = restart_at.fuse() => {}, _ = restart_at.fuse() => {},
@ -322,7 +463,7 @@ impl System {
} }
} }
async fn pull_config(self: Arc<Self>, peer: NodeID) { async fn pull_config(self: Arc<Self>, peer: Uuid) {
let resp = self let resp = self
.rpc .rpc
.call( .call(
@ -340,24 +481,14 @@ impl System {
#[async_trait] #[async_trait]
impl EndpointHandler<SystemRpc> for System { impl EndpointHandler<SystemRpc> for System {
async fn handle(self: &Arc<Self>, msg: &SystemRpc, _from: NodeID) -> SystemRpc { async fn handle(self: &Arc<Self>, msg: &SystemRpc, from: NodeID) -> Result<SystemRpc, Error> {
let resp = match msg { match msg {
SystemRpc::Connect(node) => self.handle_connect(node).await,
SystemRpc::PullConfig => Ok(self.handle_pull_config()), SystemRpc::PullConfig => Ok(self.handle_pull_config()),
SystemRpc::AdvertiseStatus(adv) => self.handle_advertise_status(from.into(), adv).await,
SystemRpc::AdvertiseConfig(adv) => self.clone().handle_advertise_config(&adv).await, SystemRpc::AdvertiseConfig(adv) => self.clone().handle_advertise_config(&adv).await,
SystemRpc::GetKnownNodes => { SystemRpc::GetKnownNodes => Ok(self.handle_get_known_nodes()),
let known_nodes = self
.fullmesh
.get_peer_list()
.iter()
.map(|n| (n.id, n.addr, n.is_up()))
.collect::<Vec<_>>();
Ok(SystemRpc::ReturnKnownNodes(known_nodes))
}
_ => Err(Error::BadRpc("Unexpected RPC message".to_string())), _ => Err(Error::BadRpc("Unexpected RPC message".to_string())),
};
match resp {
Ok(r) => r,
Err(e) => SystemRpc::Error(format!("{}", e)),
} }
} }
} }

View file

@ -36,11 +36,10 @@ enum GcRpc {
Update(Vec<ByteBuf>), Update(Vec<ByteBuf>),
DeleteIfEqualHash(Vec<(ByteBuf, Hash)>), DeleteIfEqualHash(Vec<(ByteBuf, Hash)>),
Ok, Ok,
Error(String),
} }
impl Message for GcRpc { impl Rpc for GcRpc {
type Response = GcRpc; type Response = Result<GcRpc, Error>;
} }
impl<F, R> TableGc<F, R> impl<F, R> TableGc<F, R>
@ -168,7 +167,7 @@ where
async fn try_send_and_delete( async fn try_send_and_delete(
&self, &self,
nodes: Vec<NodeID>, nodes: Vec<Uuid>,
items: Vec<(ByteBuf, Hash, ByteBuf)>, items: Vec<(ByteBuf, Hash, ByteBuf)>,
) -> Result<(), Error> { ) -> Result<(), Error> {
let n_items = items.len(); let n_items = items.len();
@ -224,8 +223,15 @@ where
.compare_and_swap::<_, _, Vec<u8>>(key, Some(vhash), None)?; .compare_and_swap::<_, _, Vec<u8>>(key, Some(vhash), None)?;
Ok(()) Ok(())
} }
}
async fn handle_rpc(&self, message: &GcRpc) -> Result<GcRpc, Error> { #[async_trait]
impl<F, R> EndpointHandler<GcRpc> for TableGc<F, R>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(self: &Arc<Self>, message: &GcRpc, _from: NodeID) -> Result<GcRpc, Error> {
match message { match message {
GcRpc::Update(items) => { GcRpc::Update(items) => {
self.data.update_many(items)?; self.data.update_many(items)?;
@ -242,16 +248,3 @@ where
} }
} }
} }
#[async_trait]
impl<F, R> EndpointHandler<GcRpc> for TableGc<F, R>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(self: &Arc<Self>, message: &GcRpc, _from: NodeID) -> GcRpc {
self.handle_rpc(message)
.await
.unwrap_or_else(|e| GcRpc::Error(format!("{}", e)))
}
}

View file

@ -2,7 +2,6 @@ use std::sync::Arc;
use garage_rpc::ring::*; use garage_rpc::ring::*;
use garage_rpc::system::System; use garage_rpc::system::System;
use garage_rpc::NodeID;
use garage_util::data::*; use garage_util::data::*;
use crate::replication::*; use crate::replication::*;
@ -20,19 +19,19 @@ pub struct TableFullReplication {
} }
impl TableReplication for TableFullReplication { impl TableReplication for TableFullReplication {
fn read_nodes(&self, _hash: &Hash) -> Vec<NodeID> { fn read_nodes(&self, _hash: &Hash) -> Vec<Uuid> {
vec![self.system.id] vec![self.system.id]
} }
fn read_quorum(&self) -> usize { fn read_quorum(&self) -> usize {
1 1
} }
fn write_nodes(&self, _hash: &Hash) -> Vec<NodeID> { fn write_nodes(&self, _hash: &Hash) -> Vec<Uuid> {
let ring = self.system.ring.borrow(); let ring = self.system.ring.borrow();
ring.config ring.config
.members .members
.keys() .keys()
.map(|id| NodeID::from_slice(id.as_slice()).unwrap()) .cloned()
.collect::<Vec<_>>() .collect::<Vec<_>>()
} }
fn write_quorum(&self) -> usize { fn write_quorum(&self) -> usize {

View file

@ -1,5 +1,4 @@
use garage_rpc::ring::*; use garage_rpc::ring::*;
use garage_rpc::NodeID;
use garage_util::data::*; use garage_util::data::*;
/// Trait to describe how a table shall be replicated /// Trait to describe how a table shall be replicated
@ -8,12 +7,12 @@ pub trait TableReplication: Send + Sync {
// To understand various replication methods // To understand various replication methods
/// Which nodes to send read requests to /// Which nodes to send read requests to
fn read_nodes(&self, hash: &Hash) -> Vec<NodeID>; fn read_nodes(&self, hash: &Hash) -> Vec<Uuid>;
/// Responses needed to consider a read succesfull /// Responses needed to consider a read succesfull
fn read_quorum(&self) -> usize; fn read_quorum(&self) -> usize;
/// Which nodes to send writes to /// Which nodes to send writes to
fn write_nodes(&self, hash: &Hash) -> Vec<NodeID>; fn write_nodes(&self, hash: &Hash) -> Vec<Uuid>;
/// Responses needed to consider a write succesfull /// Responses needed to consider a write succesfull
fn write_quorum(&self) -> usize; fn write_quorum(&self) -> usize;
fn max_write_errors(&self) -> usize; fn max_write_errors(&self) -> usize;

View file

@ -2,7 +2,6 @@ use std::sync::Arc;
use garage_rpc::ring::*; use garage_rpc::ring::*;
use garage_rpc::system::System; use garage_rpc::system::System;
use garage_rpc::NodeID;
use garage_util::data::*; use garage_util::data::*;
use crate::replication::*; use crate::replication::*;
@ -26,7 +25,7 @@ pub struct TableShardedReplication {
} }
impl TableReplication for TableShardedReplication { impl TableReplication for TableShardedReplication {
fn read_nodes(&self, hash: &Hash) -> Vec<NodeID> { fn read_nodes(&self, hash: &Hash) -> Vec<Uuid> {
let ring = self.system.ring.borrow(); let ring = self.system.ring.borrow();
ring.get_nodes(&hash, self.replication_factor) ring.get_nodes(&hash, self.replication_factor)
} }
@ -34,7 +33,7 @@ impl TableReplication for TableShardedReplication {
self.read_quorum self.read_quorum
} }
fn write_nodes(&self, hash: &Hash) -> Vec<NodeID> { fn write_nodes(&self, hash: &Hash) -> Vec<Uuid> {
let ring = self.system.ring.borrow(); let ring = self.system.ring.borrow();
ring.get_nodes(&hash, self.replication_factor) ring.get_nodes(&hash, self.replication_factor)
} }

View file

@ -45,11 +45,10 @@ pub(crate) enum SyncRpc {
Node(MerkleNodeKey, MerkleNode), Node(MerkleNodeKey, MerkleNode),
Items(Vec<Arc<ByteBuf>>), Items(Vec<Arc<ByteBuf>>),
Ok, Ok,
Error(String),
} }
impl Message for SyncRpc { impl Rpc for SyncRpc {
type Response = SyncRpc; type Response = Result<SyncRpc, Error>;
} }
struct SyncTodo { struct SyncTodo {
@ -305,7 +304,7 @@ where
async fn offload_items( async fn offload_items(
self: &Arc<Self>, self: &Arc<Self>,
items: &[(Vec<u8>, Arc<ByteBuf>)], items: &[(Vec<u8>, Arc<ByteBuf>)],
nodes: &[NodeID], nodes: &[Uuid],
) -> Result<(), Error> { ) -> Result<(), Error> {
let values = items.iter().map(|(_k, v)| v.clone()).collect::<Vec<_>>(); let values = items.iter().map(|(_k, v)| v.clone()).collect::<Vec<_>>();
@ -354,7 +353,7 @@ where
async fn do_sync_with( async fn do_sync_with(
self: Arc<Self>, self: Arc<Self>,
partition: TodoPartition, partition: TodoPartition,
who: NodeID, who: Uuid,
must_exit: watch::Receiver<bool>, must_exit: watch::Receiver<bool>,
) -> Result<(), Error> { ) -> Result<(), Error> {
let (root_ck_key, root_ck) = self.get_root_ck(partition.partition)?; let (root_ck_key, root_ck) = self.get_root_ck(partition.partition)?;
@ -480,7 +479,7 @@ where
Ok(()) Ok(())
} }
async fn send_items(&self, who: NodeID, item_value_list: Vec<Vec<u8>>) -> Result<(), Error> { async fn send_items(&self, who: Uuid, item_value_list: Vec<Vec<u8>>) -> Result<(), Error> {
info!( info!(
"({}) Sending {} items to {:?}", "({}) Sending {} items to {:?}",
self.data.name, self.data.name,
@ -513,9 +512,17 @@ where
))) )))
} }
} }
}
// ======= SYNCHRONIZATION PROCEDURE -- RECEIVER SIDE ====== // ======= SYNCHRONIZATION PROCEDURE -- RECEIVER SIDE ======
async fn handle_rpc(self: &Arc<Self>, message: &SyncRpc) -> Result<SyncRpc, Error> {
#[async_trait]
impl<F, R> EndpointHandler<SyncRpc> for TableSyncer<F, R>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(self: &Arc<Self>, message: &SyncRpc, _from: NodeID) -> Result<SyncRpc, Error> {
match message { match message {
SyncRpc::RootCkHash(range, h) => { SyncRpc::RootCkHash(range, h) => {
let (_root_ck_key, root_ck) = self.get_root_ck(*range)?; let (_root_ck_key, root_ck) = self.get_root_ck(*range)?;
@ -535,19 +542,6 @@ where
} }
} }
#[async_trait]
impl<F, R> EndpointHandler<SyncRpc> for TableSyncer<F, R>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(self: &Arc<Self>, message: &SyncRpc, _from: NodeID) -> SyncRpc {
self.handle_rpc(message)
.await
.unwrap_or_else(|e| SyncRpc::Error(format!("{}", e)))
}
}
impl SyncTodo { impl SyncTodo {
fn add_full_sync<F: TableSchema, R: TableReplication>( fn add_full_sync<F: TableSchema, R: TableReplication>(
&mut self, &mut self,

View file

@ -34,7 +34,6 @@ pub struct Table<F: TableSchema + 'static, R: TableReplication + 'static> {
#[derive(Serialize, Deserialize)] #[derive(Serialize, Deserialize)]
pub(crate) enum TableRpc<F: TableSchema> { pub(crate) enum TableRpc<F: TableSchema> {
Ok, Ok,
Error(String),
ReadEntry(F::P, F::S), ReadEntry(F::P, F::S),
ReadEntryResponse(Option<ByteBuf>), ReadEntryResponse(Option<ByteBuf>),
@ -45,8 +44,8 @@ pub(crate) enum TableRpc<F: TableSchema> {
Update(Vec<Arc<ByteBuf>>), Update(Vec<Arc<ByteBuf>>),
} }
impl<F: TableSchema> Message for TableRpc<F> { impl<F: TableSchema> Rpc for TableRpc<F> {
type Response = TableRpc<F>; type Response = Result<TableRpc<F>, Error>;
} }
impl<F, R> Table<F, R> impl<F, R> Table<F, R>
@ -277,7 +276,7 @@ where
// =============== UTILITY FUNCTION FOR CLIENT OPERATIONS =============== // =============== UTILITY FUNCTION FOR CLIENT OPERATIONS ===============
async fn repair_on_read(&self, who: &[NodeID], what: F::E) -> Result<(), Error> { async fn repair_on_read(&self, who: &[Uuid], what: F::E) -> Result<(), Error> {
let what_enc = Arc::new(ByteBuf::from(rmp_to_vec_all_named(&what)?)); let what_enc = Arc::new(ByteBuf::from(rmp_to_vec_all_named(&what)?));
self.system self.system
.rpc .rpc
@ -292,10 +291,19 @@ where
.await?; .await?;
Ok(()) Ok(())
} }
}
// ====== RPC HANDLER ===== #[async_trait]
// impl<F, R> EndpointHandler<TableRpc<F>> for Table<F, R>
async fn handle_rpc(self: &Arc<Self>, msg: &TableRpc<F>) -> Result<TableRpc<F>, Error> { where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(
self: &Arc<Self>,
msg: &TableRpc<F>,
_from: NodeID,
) -> Result<TableRpc<F>, Error> {
match msg { match msg {
TableRpc::ReadEntry(key, sort_key) => { TableRpc::ReadEntry(key, sort_key) => {
let value = self.data.read_entry(key, sort_key)?; let value = self.data.read_entry(key, sort_key)?;
@ -313,16 +321,3 @@ where
} }
} }
} }
#[async_trait]
impl<F, R> EndpointHandler<TableRpc<F>> for Table<F, R>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
async fn handle(self: &Arc<Self>, msg: &TableRpc<F>, _from: NodeID) -> TableRpc<F> {
self.handle_rpc(msg)
.await
.unwrap_or_else(|e| TableRpc::<F>::Error(format!("{}", e)))
}
}

View file

@ -7,6 +7,7 @@ use serde::de::Error as SerdeError;
use serde::{de, Deserialize}; use serde::{de, Deserialize};
use netapp::NodeID; use netapp::NodeID;
use netapp::util::parse_and_resolve_peer_addr;
use crate::error::Error; use crate::error::Error;
@ -34,11 +35,13 @@ pub struct Config {
/// Address to bind for RPC /// Address to bind for RPC
pub rpc_bind_addr: SocketAddr, pub rpc_bind_addr: SocketAddr,
/// Public IP address of this node
pub rpc_public_addr: Option<SocketAddr>,
/// Bootstrap peers RPC address /// Bootstrap peers RPC address
#[serde(deserialize_with = "deserialize_vec_addr")] #[serde(deserialize_with = "deserialize_vec_addr")]
pub bootstrap_peers: Vec<(NodeID, SocketAddr)>, pub bootstrap_peers: Vec<(NodeID, SocketAddr)>,
/// Consule host to connect to to discover more peers /// Consul host to connect to to discover more peers
pub consul_host: Option<String>, pub consul_host: Option<String>,
/// Consul service name to use /// Consul service name to use
pub consul_service_name: Option<String>, pub consul_service_name: Option<String>,
@ -111,26 +114,13 @@ fn deserialize_vec_addr<'de, D>(deserializer: D) -> Result<Vec<(NodeID, SocketAd
where where
D: de::Deserializer<'de>, D: de::Deserializer<'de>,
{ {
use std::net::ToSocketAddrs;
let mut ret = vec![]; let mut ret = vec![];
for peer in <Vec<&str>>::deserialize(deserializer)? { for peer in <Vec<&str>>::deserialize(deserializer)? {
let delim = peer let (pubkey, addrs) = parse_and_resolve_peer_addr(peer)
.find('@') .ok_or_else(|| D::Error::custom(format!("Unable to parse or resolve peer: {}", peer)))?;
.ok_or_else(|| D::Error::custom("Invalid bootstrap peer: public key not specified"))?; for ip in addrs {
let (key, host) = peer.split_at(delim); ret.push((pubkey.clone(), ip));
let pubkey = NodeID::from_slice(&hex::decode(&key).map_err(D::Error::custom)?)
.ok_or_else(|| D::Error::custom("Invalid bootstrap peer public key"))?;
let hosts = host[1..]
.to_socket_addrs()
.map_err(D::Error::custom)?
.collect::<Vec<_>>();
if hosts.is_empty() {
return Err(D::Error::custom(format!("Error resolving {}", &host[1..])));
}
for host in hosts {
ret.push((pubkey.clone(), host));
} }
} }

View file

@ -87,6 +87,18 @@ impl FixedBytes32 {
} }
} }
impl From<netapp::NodeID> for FixedBytes32 {
fn from(node_id: netapp::NodeID) -> FixedBytes32 {
FixedBytes32::try_from(node_id.as_ref()).unwrap()
}
}
impl Into<netapp::NodeID> for FixedBytes32 {
fn into(self) -> netapp::NodeID {
netapp::NodeID::from_slice(self.as_slice()).unwrap()
}
}
/// A 32 bytes UUID /// A 32 bytes UUID
pub type Uuid = FixedBytes32; pub type Uuid = FixedBytes32;
/// A 256 bit cryptographic hash, can be sha256 or blake2 depending on provenance /// A 256 bit cryptographic hash, can be sha256 or blake2 depending on provenance

View file

@ -1,35 +1,13 @@
//! Module containing error types used in Garage //! Module containing error types used in Garage
use err_derive::Error; use std::fmt;
use hyper::StatusCode;
use std::io; use std::io;
use err_derive::Error;
use serde::{de::Visitor, Deserialize, Deserializer, Serialize, Serializer};
use crate::data::*; use crate::data::*;
/// RPC related errors
#[derive(Debug, Error)]
pub enum RpcError {
#[error(display = "Node is down: {:?}.", _0)]
NodeDown(Uuid),
#[error(display = "Timeout")]
Timeout,
#[error(display = "HTTP error: {}", _0)]
Http(#[error(source)] http::Error),
#[error(display = "Hyper error: {}", _0)]
Hyper(#[error(source)] hyper::Error),
#[error(display = "Messagepack encode error: {}", _0)]
RmpEncode(#[error(source)] rmp_serde::encode::Error),
#[error(display = "Messagepack decode error: {}", _0)]
RmpDecode(#[error(source)] rmp_serde::decode::Error),
#[error(display = "Too many errors: {:?}", _0)]
TooManyErrors(Vec<String>),
}
/// Regroup all Garage errors /// Regroup all Garage errors
#[derive(Debug, Error)] #[derive(Debug, Error)]
pub enum Error { pub enum Error {
@ -63,11 +41,14 @@ pub enum Error {
#[error(display = "Tokio join error: {}", _0)] #[error(display = "Tokio join error: {}", _0)]
TokioJoin(#[error(source)] tokio::task::JoinError), TokioJoin(#[error(source)] tokio::task::JoinError),
#[error(display = "RPC call error: {}", _0)] #[error(display = "Remote error: {}", _0)]
Rpc(#[error(source)] RpcError), RemoteError(String),
#[error(display = "Remote error: {} (status code {})", _0, _1)] #[error(display = "Timeout")]
RemoteError(String, StatusCode), Timeout,
#[error(display = "Too many errors: {:?}", _0)]
TooManyErrors(Vec<String>),
#[error(display = "Bad RPC: {}", _0)] #[error(display = "Bad RPC: {}", _0)]
BadRpc(String), BadRpc(String),
@ -99,3 +80,44 @@ impl<T> From<tokio::sync::mpsc::error::SendError<T>> for Error {
Error::Message("MPSC send error".to_string()) Error::Message("MPSC send error".to_string())
} }
} }
// Custom serialization for our error type, for use in RPC.
// Errors are serialized as a string of their Display representation.
// Upon deserialization, they all become a RemoteError with the
// given representation.
impl Serialize for Error {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
serializer.serialize_str(&format!("{}", self))
}
}
impl<'de> Deserialize<'de> for Error {
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: Deserializer<'de>,
{
deserializer.deserialize_string(ErrorVisitor)
}
}
struct ErrorVisitor;
impl<'de> Visitor<'de> for ErrorVisitor {
type Value = Error;
fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
write!(formatter, "a string that represents an error value")
}
fn visit_str<E>(self, error_msg: &str) -> Result<Self::Value, E> {
Ok(Error::RemoteError(error_msg.to_string()))
}
fn visit_string<E>(self, error_msg: String) -> Result<Self::Value, E> {
Ok(Error::RemoteError(error_msg))
}
}

View file

@ -26,4 +26,4 @@ percent-encoding = "2.1.0"
futures = "0.3" futures = "0.3"
http = "0.2" http = "0.2"
hyper = "0.14" hyper = { version = "0.14", features = ["server", "http1", "runtime", "tcp", "stream"] }

View file

@ -38,7 +38,9 @@ impl Error {
match self { match self {
Error::NotFound => StatusCode::NOT_FOUND, Error::NotFound => StatusCode::NOT_FOUND,
Error::ApiError(e) => e.http_status_code(), Error::ApiError(e) => e.http_status_code(),
Error::InternalError(GarageError::Rpc(_)) => StatusCode::SERVICE_UNAVAILABLE, Error::InternalError(
GarageError::Timeout | GarageError::RemoteError(_) | GarageError::TooManyErrors(_),
) => StatusCode::SERVICE_UNAVAILABLE,
Error::InternalError(_) => StatusCode::INTERNAL_SERVER_ERROR, Error::InternalError(_) => StatusCode::INTERNAL_SERVER_ERROR,
_ => StatusCode::BAD_REQUEST, _ => StatusCode::BAD_REQUEST,
} }