hgcommands: add debugfsync

Summary:
The `debugfsync` command calls fsync on newly modified files in svfs.
Right now it only includes locations that we know have constant number
of files.

The fsync logic is put in a separate crate to avoid slow compiles.

Reviewed By: DurhamG

Differential Revision: D22992103

fbshipit-source-id: b5503e498d5216d4ba19701ecd5582387e4f45f5

eden/scm/lib/fsyncglob/Cargo.toml

@@ -0,0 +1,11 @@
+[package]
+name = "fsyncglob"
+version = "0.1.0"
+edition = "2018"
+
+[dependencies]
+glob = "0.3"
+tracing = "0.1"
+
+[dev-dependencies]
+tempfile = "3"

eden/scm/lib/fsyncglob/src/lib.rs

@@ -0,0 +1,138 @@
+/*
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This software may be used and distributed according to the terms of the
+ * GNU General Public License version 2.
+ */
+
+//! Simple crate to call fsync on files matching glob patterns.
+//!
+//! This is a standalone crate to help reducing compile time of `hgcommands`.
+
+use glob::Pattern;
+use std::fs;
+use std::io;
+use std::path::Path;
+use std::path::PathBuf;
+use std::time::Duration;
+use std::time::SystemTime;
+use tracing::debug;
+use tracing::trace;
+use tracing::warn;
+
+/// Call `fsync` on files matching given glob patterns under the given directory.
+///
+/// Errors are silenced and logged to tracing framework.
+/// Files not recently modified (older than `newer_than`) are skipped.
+///
+/// Returns paths that are fsync-ed.
+pub fn fsync_glob(dir: &Path, patterns: &[&str], newer_than: Option<SystemTime>) -> Vec<PathBuf> {
+    let escaped_dir = Pattern::escape(&dir.display().to_string());
+    let mut result = Vec::new();
+    for p in patterns {
+        let full_pattern = format!("{}/{}", &escaped_dir, p);
+        debug!("globing {}", &full_pattern);
+
+        let matches = match glob::glob(&full_pattern) {
+            Err(e) => {
+                warn!("glob failed: {}", e);
+                continue;
+            }
+            Ok(matches) => matches,
+        };
+
+        let newer_than = newer_than.unwrap_or_else(|| {
+            let now = SystemTime::now();
+            now.checked_sub(Duration::from_secs(300)).unwrap_or(now)
+        });
+
+        for path in matches {
+            let path = match path {
+                Ok(path) => path,
+                Err(e) => {
+                    warn!("path reading failed: {}", e);
+                    continue;
+                }
+            };
+
+            match try_fsync_if_newer_than(&path, newer_than) {
+                Ok(true) => {
+                    if let Ok(path) = path.strip_prefix(dir) {
+                        result.push(path.to_path_buf());
+                    }
+                    debug!("fsynced: {}", path.display());
+                }
+                Ok(false) => trace!("skipped: {}", path.display()),
+                Err(e) => warn!("cannot fsync {}: {}", path.display(), e),
+            }
+        }
+    }
+    result.sort_unstable();
+    result
+}
+
+/// Attempt to fsync a single file.
+/// Return false if the file is skipped (not newly modified or not a file).
+/// Return true if the file is synced.
+fn try_fsync_if_newer_than(path: &Path, newer_than: SystemTime) -> io::Result<bool> {
+    let metadata = path.symlink_metadata()?;
+    if !metadata.is_file() || metadata.modified()? < newer_than {
+        return Ok(false);
+    }
+
+    let mut open_opts = fs::OpenOptions::new();
+    open_opts.read(true).create(false).truncate(false);
+
+    // Windows requires opening with write permission for fsync.
+    if cfg!(windows) {
+        open_opts.write(true);
+    }
+
+    let file = open_opts.open(path)?;
+    file.sync_all()?;
+    Ok(true)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::fs;
+    use tempfile::tempdir;
+
+    #[test]
+    fn test_patterns() {
+        let dir = tempdir().unwrap();
+        let dir = dir.path();
+
+        fs::write(dir.join("a"), b"1").unwrap();
+        fs::write(dir.join("a1"), b"1").unwrap();
+        fs::write(dir.join("b"), b"2").unwrap();
+        fs::write(dir.join("c"), b"3").unwrap();
+
+        assert_eq!(d(fsync_glob(&dir, &[], None)), "[]");
+        assert_eq!(d(fsync_glob(&dir, &["d"], None)), "[]");
+        assert_eq!(d(fsync_glob(&dir, &["?"], None)), "[\"a\", \"b\", \"c\"]");
+        assert_eq!(
+            d(fsync_glob(&dir, &["a*", "c"], None)),
+            "[\"a\", \"a1\", \"c\"]"
+        );
+    }
+
+    #[test]
+    fn test_skip_old_files() {
+        let dir = tempdir().unwrap();
+        let dir = dir.path();
+
+        fs::write(dir.join("a"), b"1").unwrap();
+        fs::write(dir.join("b"), b"2").unwrap();
+
+        let newer_than = SystemTime::now()
+            .checked_add(Duration::from_secs(10))
+            .unwrap();
+        assert_eq!(d(fsync_glob(&dir, &["*"], Some(newer_than))), "[]");
+    }
+
+    fn d(value: impl std::fmt::Debug) -> String {
+        format!("{:?}", value)
+    }
+}

eden/scm/lib/hgcommands/Cargo.toml

@@ -24,6 +24,7 @@ encoding = { path = "../encoding" }
 env_logger = "0.7"
 filetime = "0.2.9"
 flate2 = "1"
+fsyncglob = { path = "../fsyncglob" }
 hgtime = { path = "../hgtime"}
 indexedlog = { path = "../indexedlog" }
 libc = "0.2"

eden/scm/lib/hgcommands/src/commands.rs

@@ -43,6 +43,7 @@ pub fn table() -> CommandTable {
         debug::dumpindexedlog,
         debug::dumptrace,
         debug::dynamicconfig,
+        debug::fsync,
         debug::http,
         debug::python,
         debug::store,

eden/scm/lib/hgcommands/src/commands/debug.rs

@@ -16,6 +16,7 @@ pub mod causerusterror;
 pub mod dumpindexedlog;
 pub mod dumptrace;
 pub mod dynamicconfig;
+pub mod fsync;
 pub mod http;
 pub mod python;
 pub mod store;

eden/scm/lib/hgcommands/src/commands/debug/fsync.rs

@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This software may be used and distributed according to the terms of the
+ * GNU General Public License version 2.
+ */
+
+use super::NoOpts;
+use super::Repo;
+use super::Result;
+use super::IO;
+
+pub fn run(_opts: NoOpts, _io: &mut IO, repo: Repo) -> Result<u8> {
+    let store_path = repo.store_path();
+    let patterns = [
+        "00changelog.*",
+        "hgcommits/**/*",
+        "metalog/**/*",
+        "mutation/**/*",
+    ];
+    fsyncglob::fsync_glob(store_path, &patterns, None);
+    Ok(0)
+}
+
+pub fn name() -> &'static str {
+    "debugfsync"
+}
+
+pub fn doc() -> &'static str {
+    "call fsync on newly modified key storage files"
+}

eden/scm/tests/test-completion.t

@@ -120,6 +120,7 @@ Show debug commands if there are no other candidates
   debugfileset
   debugformat
   debugfsinfo
+  debugfsync
   debuggetbundle
   debughttp
   debugignore
@@ -401,6 +402,7 @@ Show all commands + options
   debugfileset: rev
   debugformat: template
   debugfsinfo: 
+  debugfsync: 
   debuggetbundle: head, common, type
   debughttp: 
   debugignore: 

eden/scm/tests/test-help.t

@@ -992,6 +992,7 @@ Test list of internal help commands
    debugfileset  parse and apply a fileset specification
    debugformat   display format information about the current repository
    debugfsinfo   show information detected about current filesystem
+   debugfsync    call fsync on newly modified key storage files
    debuggentrees
                  (no help text available)
    debuggetbundle