itsi-scheduler 0.1.5 → 0.1.19

data/ext/itsi_server/src/ruby_types/itsi_server/file_watcher.rs

@@ -0,0 +1,225 @@
+use derive_more::Debug;
+use globset::{Glob, GlobSet, GlobSetBuilder};
+use magnus::error::Result;
+use nix::unistd::{close, fork, pipe, read};
+use notify::{event::ModifyKind, EventKind, RecommendedWatcher};
+use notify::{Event, RecursiveMode, Watcher};
+use std::path::Path;
+use std::sync::mpsc::Sender;
+use std::time::{Duration, Instant};
+use std::{collections::HashSet, fs};
+use std::{
+    os::fd::{AsRawFd, IntoRawFd, OwnedFd},
+    path::PathBuf,
+    process::Command,
+    sync::mpsc,
+    thread::{self},
+};
+
+/// Represents a set of patterns and commands.
+#[derive(Debug, Clone)]
+struct PatternGroup {
+    base_dir: PathBuf,
+    glob_set: GlobSet,
+    pattern: String,
+    commands: Vec<Vec<String>>,
+    last_triggered: Option<Instant>,
+}
+
+/// Extracts the base directory from a wildcard pattern by taking the portion up to the first
+/// component that contains a wildcard character.
+fn extract_and_canonicalize_base_dir(pattern: &str) -> PathBuf {
+    if !(pattern.contains("*") || pattern.contains("?") || pattern.contains('[')) {
+        if let Ok(metadata) = fs::metadata(pattern) {
+            if metadata.is_dir() {
+                return fs::canonicalize(pattern).unwrap();
+            }
+            if metadata.is_file() {
+                return fs::canonicalize(pattern)
+                    .unwrap()
+                    .parent()
+                    .unwrap()
+                    .to_path_buf();
+            }
+        }
+    }
+
+    let path = Path::new(pattern);
+    let mut base = PathBuf::new();
+    for comp in path.components() {
+        let comp_str = comp.as_os_str().to_string_lossy();
+        if comp_str.contains('*') || comp_str.contains('?') || comp_str.contains('[') {
+            break;
+        } else {
+            base.push(comp);
+        }
+    }
+    // If no base was built, default to "."
+    let base = if base.as_os_str().is_empty() || !base.exists() {
+        PathBuf::from(".")
+    } else {
+        base
+    };
+
+    // Canonicalize to get the absolute path.
+    fs::canonicalize(&base).unwrap_or(base)
+}
+
+/// Minimum time between triggering the same pattern group (debounce time)
+const DEBOUNCE_DURATION: Duration = Duration::from_millis(500);
+
+pub fn watch_groups(pattern_groups: Vec<(String, Vec<Vec<String>>)>) -> Result<Option<OwnedFd>> {
+    let (r_fd, w_fd): (OwnedFd, OwnedFd) = pipe().map_err(|e| {
+        magnus::Error::new(
+            magnus::exception::standard_error(),
+            format!("Failed to create watcher pipe: {}", e),
+        )
+    })?;
+
+    let fork_result = unsafe {
+        fork().map_err(|e| {
+            magnus::Error::new(
+                magnus::exception::standard_error(),
+                format!("Failed to fork file watcher: {}", e),
+            )
+        })
+    }?;
+
+    if fork_result.is_child() {
+        let _ = close(w_fd.into_raw_fd());
+        thread::spawn(move || {
+            let mut buf = [0u8; 1];
+            loop {
+                match read(r_fd.as_raw_fd(), &mut buf) {
+                    Ok(0) => {
+                        std::process::exit(0);
+                    }
+                    Ok(_) => {}
+                    Err(_) => {
+                        std::process::exit(0);
+                    }
+                }
+            }
+        });
+
+        let mut groups = Vec::new();
+        for (pattern, commands) in pattern_groups.into_iter() {
+            let base_dir = extract_and_canonicalize_base_dir(&pattern);
+            let glob = Glob::new(&pattern).map_err(|e| {
+                magnus::Error::new(
+                    magnus::exception::standard_error(),
+                    format!("Failed to create watch glob: {}", e),
+                )
+            })?;
+            let glob_set = GlobSetBuilder::new().add(glob).build().map_err(|e| {
+                magnus::Error::new(
+                    magnus::exception::standard_error(),
+                    format!("Failed to create watch glob set: {}", e),
+                )
+            })?;
+            groups.push(PatternGroup {
+                base_dir,
+                glob_set,
+                pattern,
+                commands,
+                last_triggered: None,
+            });
+        }
+
+        // Create a channel and a watcher.
+        let (tx, rx) = mpsc::channel::<notify::Result<Event>>();
+        let sender = tx.clone();
+        fn event_fn(sender: Sender<notify::Result<Event>>) -> impl Fn(notify::Result<Event>) {
+            move |res| match res {
+                Ok(event) => {
+                    sender.send(Ok(event)).unwrap();
+                }
+                Err(e) => println!("watch error: {:?}", e),
+            }
+        }
+
+        let mut watched_dirs = HashSet::new();
+        let mut watcher: RecommendedWatcher =
+            notify::recommended_watcher(event_fn(sender)).expect("Failed to create watcher");
+        for group in &groups {
+            if watched_dirs.insert(group.base_dir.clone()) {
+                watcher
+                    .watch(&group.base_dir, RecursiveMode::Recursive)
+                    .expect("Failed to add watch");
+            }
+        }
+
+        // Main event loop.
+        for res in rx {
+            match res {
+                Ok(event) => {
+                    if !matches!(event.kind, EventKind::Modify(ModifyKind::Metadata(_))) {
+                        continue;
+                    }
+                    let now = Instant::now();
+                    for group in &mut groups {
+                        for path in event.paths.iter() {
+                            if let Ok(rel_path) = path.strip_prefix(&group.base_dir) {
+                                if group.glob_set.is_match(rel_path)
+                                    || rel_path.to_str().is_some_and(|s| s == group.pattern)
+                                {
+                                    // Check if we should debounce this event
+                                    if let Some(last_triggered) = group.last_triggered {
+                                        if now.duration_since(last_triggered) < DEBOUNCE_DURATION {
+                                            // Skip this event as we've recently triggered for this pattern
+                                            continue;
+                                        }
+                                    }
+
+                                    // Update the last triggered time
+                                    group.last_triggered = Some(now);
+
+                                    // Execute the commands for this group.
+                                    for command in &group.commands {
+                                        if command.is_empty() {
+                                            continue;
+                                        }
+                                        let mut cmd = Command::new(&command[0]);
+                                        if command.len() > 1 {
+                                            cmd.args(&command[1..]);
+                                        }
+                                        match cmd.spawn() {
+                                            Ok(mut child) => {
+                                                if let Err(e) = child.wait() {
+                                                    eprintln!(
+                                                        "Command {:?} failed: {:?}",
+                                                        command, e
+                                                    );
+                                                }
+                                            }
+                                            Err(e) => {
+                                                eprintln!(
+                                                    "Failed to execute command {:?}: {:?}",
+                                                    command, e
+                                                );
+                                            }
+                                        }
+                                    }
+                                    break;
+                                }
+                            }
+                        }
+                    }
+                }
+                Err(e) => println!("Watch error: {:?}", e),
+            }
+        }
+
+        // Clean up the watches.
+        for group in &groups {
+            watcher
+                .unwatch(&group.base_dir)
+                .expect("Failed to remove watch");
+        }
+        drop(watcher);
+        std::process::exit(0);
+    } else {
+        let _ = close(r_fd.into_raw_fd());
+        Ok(Some(w_fd))
+    }
+}

data/ext/itsi_server/src/ruby_types/itsi_server/itsi_server_config.rs

@@ -0,0 +1,375 @@
+use super::file_watcher::{self};
+use crate::{
+    ruby_types::ITSI_SERVER_CONFIG,
+    server::{
+        binds::{bind::Bind, listener::Listener},
+        middleware_stack::MiddlewareSet,
+    },
+};
+use derive_more::Debug;
+use itsi_rb_helpers::{call_with_gvl, print_rb_backtrace, HeapValue};
+use itsi_tracing::{set_format, set_level, set_target};
+use magnus::{
+    block::Proc,
+    error::Result,
+    value::{LazyId, ReprValue},
+    RArray, RHash, Ruby, Symbol, Value,
+};
+use nix::{
+    fcntl::{fcntl, FcntlArg, FdFlag},
+    unistd::{close, dup},
+};
+use parking_lot::{Mutex, RwLock};
+use std::{
+    collections::HashMap,
+    os::fd::{AsRawFd, OwnedFd, RawFd},
+    path::PathBuf,
+    sync::{Arc, OnceLock},
+    time::Duration,
+};
+static DEFAULT_BIND: &str = "http://localhost:3000";
+static ID_BUILD_CONFIG: LazyId = LazyId::new("build_config");
+static ID_RELOAD_EXEC: LazyId = LazyId::new("reload_exec");
+
+#[derive(Debug, Clone)]
+pub struct ItsiServerConfig {
+    pub cli_params: Arc<HeapValue<RHash>>,
+    pub itsifile_path: Option<PathBuf>,
+    pub itsi_config_proc: Arc<Option<HeapValue<Proc>>>,
+    #[debug(skip)]
+    pub server_params: Arc<RwLock<Arc<ServerParams>>>,
+    pub watcher_fd: Arc<Option<OwnedFd>>,
+}
+
+#[derive(Debug)]
+pub struct ServerParams {
+    /// Cluster params
+    pub workers: u8,
+    pub worker_memory_limit: Option<u64>,
+    pub silence: bool,
+    pub shutdown_timeout: f64,
+    pub hooks: HashMap<String, HeapValue<Proc>>,
+    pub preload: bool,
+
+    pub request_timeout: Option<Duration>,
+    pub notify_watchers: Option<Vec<(String, Vec<Vec<String>>)>>,
+    /// Worker params
+    pub threads: u8,
+    pub scheduler_threads: Option<u8>,
+    pub streamable_body: bool,
+    pub multithreaded_reactor: bool,
+    pub pin_worker_cores: bool,
+    pub scheduler_class: Option<String>,
+    pub oob_gc_responses_threshold: Option<u64>,
+    pub middleware_loader: HeapValue<Proc>,
+    pub middleware: OnceLock<MiddlewareSet>,
+    pub binds: Vec<Bind>,
+    #[debug(skip)]
+    pub(crate) listeners: Mutex<Vec<Listener>>,
+    listener_info: Mutex<HashMap<String, i32>>,
+}
+
+impl ServerParams {
+    pub fn preload_ruby(self: &Arc<Self>) -> Result<()> {
+        call_with_gvl(|ruby| -> Result<()> {
+            if self
+                .scheduler_class
+                .as_ref()
+                .is_some_and(|t| t == "Itsi::Scheduler")
+            {
+                ruby.require("itsi/scheduler")?;
+            }
+            let middleware = MiddlewareSet::new(
+                self.middleware_loader
+                    .call::<_, Option<Value>>(())
+                    .inspect_err(|e| {
+                        if let Some(err_value) = e.value() {
+                            print_rb_backtrace(err_value);
+                        }
+                    })?
+                    .map(|mw| mw.into()),
+            )?;
+            self.middleware.set(middleware).map_err(|_| {
+                magnus::Error::new(
+                    magnus::exception::runtime_error(),
+                    "Failed to set middleware",
+                )
+            })?;
+            Ok(())
+        })?;
+        Ok(())
+    }
+
+    fn from_rb_hash(rb_param_hash: RHash) -> Result<ServerParams> {
+        let workers = rb_param_hash
+            .fetch::<_, Option<u8>>("workers")?
+            .unwrap_or(num_cpus::get() as u8);
+        let worker_memory_limit: Option<u64> = rb_param_hash.fetch("worker_memory_limit")?;
+        let silence: bool = rb_param_hash.fetch("silence")?;
+        let multithreaded_reactor: bool = rb_param_hash
+            .fetch::<_, Option<bool>>("multithreaded_reactor")?
+            .unwrap_or(workers == 1);
+        let pin_worker_cores: bool = rb_param_hash
+            .fetch::<_, Option<bool>>("pin_worker_cores")?
+            .unwrap_or(true);
+        let shutdown_timeout: f64 = rb_param_hash.fetch("shutdown_timeout")?;
+
+        let hooks: Option<RHash> = rb_param_hash.fetch("hooks")?;
+        let hooks = hooks
+            .map(|rhash| -> Result<HashMap<String, HeapValue<Proc>>> {
+                let mut hook_map: HashMap<String, HeapValue<Proc>> = HashMap::new();
+                for pair in rhash.enumeratorize::<_, ()>("each", ()) {
+                    if let Some(pair_value) = RArray::from_value(pair?) {
+                        if let (Ok(key), Ok(value)) =
+                            (pair_value.entry::<Value>(0), pair_value.entry::<Proc>(1))
+                        {
+                            hook_map.insert(key.to_string(), HeapValue::from(value));
+                        }
+                    }
+                }
+                Ok(hook_map)
+            })
+            .transpose()?
+            .unwrap_or_default();
+        let preload: bool = rb_param_hash.fetch("preload")?;
+        let request_timeout: Option<u64> = rb_param_hash.fetch("request_timeout")?;
+        let request_timeout = request_timeout.map(Duration::from_secs);
+
+        let notify_watchers: Option<Vec<(String, Vec<Vec<String>>)>> =
+            rb_param_hash.fetch("notify_watchers")?;
+        let threads: u8 = rb_param_hash.fetch("threads")?;
+        let scheduler_threads: Option<u8> = rb_param_hash.fetch("scheduler_threads")?;
+        let streamable_body: bool = rb_param_hash.fetch("streamable_body")?;
+        let scheduler_class: Option<String> = rb_param_hash.fetch("scheduler_class")?;
+        let oob_gc_responses_threshold: Option<u64> =
+            rb_param_hash.fetch("oob_gc_responses_threshold")?;
+        let middleware_loader: Proc = rb_param_hash.fetch("middleware_loader")?;
+        let log_level: Option<String> = rb_param_hash.fetch("log_level")?;
+        let log_target: Option<String> = rb_param_hash.fetch("log_target")?;
+        let log_format: Option<String> = rb_param_hash.fetch("log_format")?;
+
+        if let Some(level) = log_level {
+            set_level(&level);
+        }
+
+        if let Some(target) = log_target {
+            set_target(&target);
+        }
+
+        if let Some(format) = log_format {
+            set_format(&format);
+        }
+
+        let binds: Option<Vec<String>> = rb_param_hash.fetch("binds")?;
+        let binds = binds
+            .unwrap_or_else(|| vec![DEFAULT_BIND.to_string()])
+            .into_iter()
+            .map(|s| s.parse())
+            .collect::<itsi_error::Result<Vec<Bind>>>()?;
+
+        let listeners = if let Some(preexisting_listeners) =
+            rb_param_hash.delete::<_, Option<String>>("listeners")?
+        {
+            let bind_to_fd_map: HashMap<String, i32> = serde_json::from_str(&preexisting_listeners)
+                .map_err(|e| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Invalid listener info: {}", e),
+                    )
+                })?;
+
+            binds
+                .iter()
+                .cloned()
+                .map(|bind| {
+                    if let Some(fd) = bind_to_fd_map.get(&bind.listener_address_string()) {
+                        Listener::inherit_fd(bind, *fd)
+                    } else {
+                        Listener::try_from(bind)
+                    }
+                })
+                .collect::<std::result::Result<Vec<Listener>, _>>()?
+                .into_iter()
+                .collect::<Vec<_>>()
+        } else {
+            binds
+                .iter()
+                .cloned()
+                .map(Listener::try_from)
+                .collect::<std::result::Result<Vec<Listener>, _>>()?
+                .into_iter()
+                .collect::<Vec<_>>()
+        };
+
+        let listener_info = listeners
+            .iter()
+            .map(|listener| {
+                listener.handover().map_err(|e| {
+                    magnus::Error::new(magnus::exception::runtime_error(), e.to_string())
+                })
+            })
+            .collect::<Result<HashMap<String, i32>>>()?;
+
+        Ok(ServerParams {
+            workers,
+            worker_memory_limit,
+            silence,
+            multithreaded_reactor,
+            pin_worker_cores,
+            shutdown_timeout,
+            hooks,
+            preload,
+            request_timeout,
+            notify_watchers,
+            threads,
+            scheduler_threads,
+            streamable_body,
+            scheduler_class,
+            oob_gc_responses_threshold,
+            binds,
+            listener_info: Mutex::new(listener_info),
+            listeners: Mutex::new(listeners),
+            middleware_loader: middleware_loader.into(),
+            middleware: OnceLock::new(),
+        })
+    }
+}
+
+impl ItsiServerConfig {
+    pub fn new(
+        ruby: &Ruby,
+        cli_params: RHash,
+        itsifile_path: Option<PathBuf>,
+        itsi_config_proc: Option<Proc>,
+    ) -> Result<Self> {
+        let itsi_config_proc = Arc::new(itsi_config_proc.map(HeapValue::from));
+        let server_params = Self::combine_params(
+            ruby,
+            cli_params,
+            itsifile_path.as_ref(),
+            itsi_config_proc.clone(),
+        )?;
+        cli_params.delete::<_, Value>(Symbol::new("listeners"))?;
+
+        let watcher_fd = if let Some(watchers) = server_params.notify_watchers.clone() {
+            file_watcher::watch_groups(watchers)?
+        } else {
+            None
+        };
+
+        Ok(ItsiServerConfig {
+            cli_params: Arc::new(cli_params.into()),
+            server_params: RwLock::new(server_params.clone()).into(),
+            itsi_config_proc,
+            itsifile_path,
+            watcher_fd: watcher_fd.into(),
+        })
+    }
+
+    /// Reload
+    pub fn reload(self: Arc<Self>, cluster_worker: bool) -> Result<bool> {
+        let server_params = call_with_gvl(|ruby| {
+            Self::combine_params(
+                &ruby,
+                self.cli_params.cloned(),
+                self.itsifile_path.as_ref(),
+                self.itsi_config_proc.clone(),
+            )
+        })?;
+
+        let is_single_mode = self.server_params.read().workers == 1;
+
+        let requires_exec = if !is_single_mode && !server_params.preload {
+            // In cluster mode children are cycled during a reload
+            // and if preload is disabled, will get a clean memory slate,
+            // so we don't need to exec.
+            false
+        } else {
+            // In non-cluster mode, or when preloading is enabled, we shouldn't try to
+            // reload inside the existing process (as new code may conflict with old),
+            // and should re-exec instead.
+            true
+        };
+
+        *self.server_params.write() = server_params.clone();
+        Ok(requires_exec && (cluster_worker || is_single_mode))
+    }
+
+    fn combine_params(
+        ruby: &Ruby,
+        cli_params: RHash,
+        itsifile_path: Option<&PathBuf>,
+        itsi_config_proc: Arc<Option<HeapValue<Proc>>>,
+    ) -> Result<Arc<ServerParams>> {
+        let inner = itsi_config_proc
+            .as_ref()
+            .clone()
+            .map(|hv| hv.clone().inner());
+        let rb_param_hash: RHash = ruby.get_inner_ref(&ITSI_SERVER_CONFIG).funcall(
+            *ID_BUILD_CONFIG,
+            (cli_params, itsifile_path.cloned(), inner),
+        )?;
+        Ok(Arc::new(ServerParams::from_rb_hash(rb_param_hash)?))
+    }
+
+    fn clear_cloexec(fd: RawFd) -> nix::Result<()> {
+        let current_flags = fcntl(fd, FcntlArg::F_GETFD)?;
+        let mut flags = FdFlag::from_bits_truncate(current_flags);
+        // Remove the FD_CLOEXEC flag
+        flags.remove(FdFlag::FD_CLOEXEC);
+        // Set the new flags back on the file descriptor
+        fcntl(fd, FcntlArg::F_SETFD(flags))?;
+        Ok(())
+    }
+
+    pub fn dup_fds(self: &Arc<Self>) -> Result<()> {
+        let binding = self.server_params.read();
+        let mut listener_info_guard = binding.listener_info.lock();
+        let dupped_fd_map = listener_info_guard
+            .iter()
+            .map(|(str, fd)| {
+                let dupped_fd = dup(*fd).map_err(|errno| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Errno {} while trying to dup {}", errno, fd),
+                    )
+                })?;
+                Self::clear_cloexec(dupped_fd).map_err(|e| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Failed to clear cloexec flag for fd {}: {}", dupped_fd, e),
+                    )
+                })?;
+                Ok((str.clone(), dupped_fd))
+            })
+            .collect::<Result<HashMap<String, i32>>>()?;
+        *listener_info_guard = dupped_fd_map;
+        Ok(())
+    }
+
+    pub fn stop_watcher(self: &Arc<Self>) -> Result<()> {
+        if let Some(r_fd) = self.watcher_fd.as_ref() {
+            close(r_fd.as_raw_fd()).ok();
+        }
+        Ok(())
+    }
+
+    pub fn reload_exec(self: &Arc<Self>) -> Result<()> {
+        let listener_json =
+            serde_json::to_string(&self.server_params.read().listener_info.lock().clone())
+                .map_err(|e| {
+                    magnus::Error::new(
+                        magnus::exception::standard_error(),
+                        format!("Invalid listener info: {}", e),
+                    )
+                })?;
+
+        self.stop_watcher()?;
+        call_with_gvl(|ruby| -> Result<()> {
+            ruby.get_inner_ref(&ITSI_SERVER_CONFIG)
+                .funcall::<_, _, Value>(*ID_RELOAD_EXEC, (listener_json,))?;
+            Ok(())
+        })?;
+        Ok(())
+    }
+}

data/ext/itsi_server/src/ruby_types/itsi_server.rs

@@ -0,0 +1,83 @@
+use crate::server::{
+    lifecycle_event::LifecycleEvent,
+    serve_strategy::{cluster_mode::ClusterMode, single_mode::SingleMode, ServeStrategy},
+    signal::{clear_signal_handlers, reset_signal_handlers, send_lifecycle_event},
+};
+use itsi_rb_helpers::{call_without_gvl, print_rb_backtrace};
+use itsi_server_config::ItsiServerConfig;
+use itsi_tracing::{error, run_silently};
+use magnus::{block::Proc, error::Result, RHash, Ruby};
+use parking_lot::Mutex;
+use std::{path::PathBuf, sync::Arc};
+use tracing::{info, instrument};
+mod file_watcher;
+pub mod itsi_server_config;
+#[magnus::wrap(class = "Itsi::Server", free_immediately, size)]
+#[derive(Clone)]
+pub struct ItsiServer {
+    pub config: Arc<Mutex<Arc<ItsiServerConfig>>>,
+}
+
+impl ItsiServer {
+    pub fn new(
+        ruby: &Ruby,
+        cli_params: RHash,
+        itsifile_path: Option<PathBuf>,
+        itsi_config_proc: Option<Proc>,
+    ) -> Result<Self> {
+        Ok(Self {
+            config: Arc::new(Mutex::new(Arc::new(ItsiServerConfig::new(
+                ruby,
+                cli_params,
+                itsifile_path,
+                itsi_config_proc,
+            )?))),
+        })
+    }
+
+    pub fn stop(&self) -> Result<()> {
+        send_lifecycle_event(LifecycleEvent::Shutdown);
+        Ok(())
+    }
+
+    #[instrument(skip(self))]
+    pub fn start(&self) -> Result<()> {
+        let result = if self.config.lock().server_params.read().silence {
+            run_silently(|| self.build_and_run_strategy())
+        } else {
+            info!("Itsi - Rolling into action. ⚪💨");
+            self.build_and_run_strategy()
+        };
+        if let Err(e) = result {
+            if let Some(err_value) = e.value() {
+                print_rb_backtrace(err_value);
+            }
+            return Err(e);
+        }
+        Ok(())
+    }
+
+    pub(crate) fn build_strategy(&self) -> Result<ServeStrategy> {
+        let server_config = self.config.lock();
+        Ok(if server_config.server_params.read().workers > 1 {
+            ServeStrategy::Cluster(Arc::new(ClusterMode::new(server_config.clone())))
+        } else {
+            ServeStrategy::Single(Arc::new(SingleMode::new(server_config.clone())?))
+        })
+    }
+
+    fn build_and_run_strategy(&self) -> Result<()> {
+        reset_signal_handlers();
+        call_without_gvl(move || -> Result<()> {
+            let strategy = self.build_strategy()?;
+            if let Err(e) = strategy.clone().run() {
+                error!("Error running server: {}", e);
+                strategy.stop()?;
+            }
+            Ok(())
+        })?;
+        clear_signal_handlers();
+        info!("Server stopped");
+        Ok(())
+    }
+}