fix: pgdump-cluster (#68)

* feat: add as_str/from_str to PostgresDumpFormat

* feat: resolve pg_dumpall/psql binary names

* feat: add include_globals field to database config

* feat: add pg_dumpall/psql globals dump and apply

* feat: add postgres backup bundle (manifest + build + resolve)

* feat: bundle globals into postgres backup when include_globals is set

* feat: replay globals before pg_restore when backup archive is a bundle

* refactor: bind FD restore tempdir guard once to clear unused warnings

* docs: demonstrate include_globals in sample databases.json

* chore: silence test-only re-export warning in non-test builds

* revert: remove include_globals feature, restore plain pg_dump/pg_restore

* feat: add pg_dumpall/psql binary names and is_superuser check

* feat: add postgresql-cluster db type and config parsing

* feat: pg_dumpall cluster backup and psql restore

* feat: route postgresql-cluster through PostgresClusterDatabase

* docs: add postgresql-cluster sample to databases.json

* refactor: split cluster mode into cluster/ module (backup, restore, database)

* test: mirror cluster tests into src/tests/domain/cluster/
This commit is contained in:
Charles GTE
2026-06-27 17:23:57 +02:00
committed by GitHub
parent 0b7b7ef59b
commit 01f5d34c96
17 changed files with 602 additions and 29 deletions
+3
View File
@@ -1,5 +1,6 @@
use crate::domain::mongodb::database::MongoDatabase;
use crate::domain::mysql::database::MySQLDatabase;
use crate::domain::postgres::cluster::database::PostgresClusterDatabase;
use crate::domain::postgres::database::PostgresDatabase;
use crate::domain::postgres::{detect_format_from_file, detect_format_from_size};
use crate::domain::redis::database::RedisDatabase;
@@ -31,6 +32,7 @@ impl DatabaseFactory {
let format = detect_format_from_size(&cfg).await;
Arc::new(PostgresDatabase::new(cfg, format))
}
DbType::PostgresqlCluster => Arc::new(PostgresClusterDatabase::new(cfg)),
DbType::Mysql => Arc::new(MySQLDatabase::new(cfg)),
DbType::Mariadb => Arc::new(MariaDBDatabase::new(cfg)),
DbType::MongoDB => Arc::new(MongoDatabase::new(cfg)),
@@ -48,6 +50,7 @@ impl DatabaseFactory {
let format = detect_format_from_file(restore_file);
Arc::new(PostgresDatabase::new(cfg, format))
}
DbType::PostgresqlCluster => Arc::new(PostgresClusterDatabase::new(cfg)),
DbType::Mysql => Arc::new(MySQLDatabase::new(cfg)),
DbType::Mariadb => Arc::new(MariaDBDatabase::new(cfg)),
DbType::MongoDB => Arc::new(MongoDatabase::new(cfg)),
+79
View File
@@ -0,0 +1,79 @@
use anyhow::Result;
use std::collections::HashMap;
use std::path::PathBuf;
use std::process::Command;
use std::sync::Arc;
use std::time::Instant;
use super::super::connection::{
is_superuser, pg_dumpall_binary_name, select_pg_path, server_version,
};
use crate::services::backup::logger::JobLogger;
use crate::services::config::DatabaseConfig;
pub async fn run(
cfg: DatabaseConfig,
backup_dir: PathBuf,
env: HashMap<String, String>,
logger: Arc<JobLogger>,
) -> Result<PathBuf> {
tokio::task::spawn_blocking(move || -> Result<PathBuf> {
logger.log("info", format!("Starting cluster backup for {}", cfg.name));
let version = match futures::executor::block_on(server_version(&cfg)) {
Ok(v) => v,
Err(e) => {
logger.log("error", format!("Failed to get server version for {}: {:?}", cfg.name, e));
return Err(e.into());
}
};
match futures::executor::block_on(is_superuser(&cfg)) {
Ok(true) => {}
Ok(false) => {
logger.log("error", format!("postgresql-cluster backup requires a superuser role for {}", cfg.name));
anyhow::bail!("postgresql-cluster backup requires a superuser role for {}", cfg.name);
}
Err(e) => {
logger.log("error", format!("Failed to check superuser status for {}: {:?}", cfg.name, e));
return Err(e.into());
}
}
let pg_dumpall = select_pg_path(&version).join(pg_dumpall_binary_name());
let file_path = backup_dir.join(format!("{}.sql", cfg.generated_id));
logger.log("info", format!("Running pg_dumpall for cluster {} via {:?}", cfg.name, pg_dumpall));
let start = Instant::now();
let output = Command::new(&pg_dumpall)
.arg("--host").arg(&cfg.host)
.arg("--port").arg(cfg.port.to_string())
.arg("--username").arg(&cfg.username)
.arg("-v")
.arg("-f").arg(&file_path)
.envs(env)
.output();
let duration_ms = start.elapsed().as_millis() as f64;
match output {
Ok(o) => {
let stderr = String::from_utf8_lossy(&o.stderr).to_string();
let exit_code = o.status.code().unwrap_or(-1);
if o.status.success() {
logger.log_command("pg_dumpall", if stderr.is_empty() { None } else { Some(stderr) }, Some(0), Some(duration_ms));
logger.log("info", format!("Cluster backup completed for {} at {:?}", cfg.name, file_path));
Ok(file_path)
} else {
logger.log_command("pg_dumpall", Some(stderr), Some(exit_code), Some(duration_ms));
anyhow::bail!("Cluster backup (pg_dumpall) failed for {}", cfg.name);
}
}
Err(e) => {
logger.log_command("pg_dumpall", Some(e.to_string()), Some(-1), Some(duration_ms));
Err(e.into())
}
}
})
.await?
}
+53
View File
@@ -0,0 +1,53 @@
use anyhow::Result;
use async_trait::async_trait;
use std::collections::HashMap;
use std::path::{Path, PathBuf};
use std::sync::Arc;
use super::super::ping;
use super::{backup, restore};
use crate::domain::factory::Database;
use crate::services::backup::logger::JobLogger;
use crate::services::config::DatabaseConfig;
use crate::utils::locks::{DbOpLock, FileLock};
pub struct PostgresClusterDatabase {
pub cfg: DatabaseConfig,
}
impl PostgresClusterDatabase {
pub fn new(cfg: DatabaseConfig) -> Self {
Self { cfg }
}
fn build_env(&self) -> HashMap<String, String> {
let mut envs = std::env::vars().collect::<HashMap<_, _>>();
envs.insert("PGPASSWORD".to_string(), self.cfg.password.to_string());
envs
}
}
#[async_trait]
impl Database for PostgresClusterDatabase {
fn file_extension(&self) -> &'static str {
".sql"
}
async fn ping(&self) -> Result<bool> {
ping::run(self.cfg.clone()).await
}
async fn backup(&self, dir: &Path, logger: Arc<JobLogger>) -> Result<PathBuf> {
FileLock::acquire(&self.cfg.generated_id, DbOpLock::Backup.as_str()).await?;
let res = backup::run(self.cfg.clone(), dir.to_path_buf(), self.build_env(), logger).await;
FileLock::release(&self.cfg.generated_id).await?;
res
}
async fn restore(&self, file: &Path, logger: Arc<JobLogger>) -> Result<()> {
FileLock::acquire(&self.cfg.generated_id, DbOpLock::Restore.as_str()).await?;
let res = restore::run(self.cfg.clone(), file.to_path_buf(), self.build_env(), logger).await;
FileLock::release(&self.cfg.generated_id).await?;
res
}
}
+3
View File
@@ -0,0 +1,3 @@
pub mod backup;
pub mod database;
pub mod restore;
+78
View File
@@ -0,0 +1,78 @@
use anyhow::Result;
use std::collections::HashMap;
use std::path::PathBuf;
use std::process::Command;
use std::sync::Arc;
use std::time::Instant;
use super::super::connection::{is_superuser, psql_binary_name, select_pg_path, server_version};
use crate::services::backup::logger::JobLogger;
use crate::services::config::DatabaseConfig;
pub async fn run(
cfg: DatabaseConfig,
restore_file: PathBuf,
env: HashMap<String, String>,
logger: Arc<JobLogger>,
) -> Result<()> {
tokio::task::spawn_blocking(move || -> Result<()> {
logger.log("info", format!("Starting cluster restore for {}", cfg.name));
let version = match futures::executor::block_on(server_version(&cfg)) {
Ok(v) => v,
Err(e) => {
logger.log("error", format!("Failed to get server version for {}: {:?}", cfg.name, e));
return Err(e.into());
}
};
match futures::executor::block_on(is_superuser(&cfg)) {
Ok(true) => {}
Ok(false) => {
logger.log("error", format!("postgresql-cluster restore requires a superuser role for {}", cfg.name));
anyhow::bail!("postgresql-cluster restore requires a superuser role for {}", cfg.name);
}
Err(e) => {
logger.log("error", format!("Failed to check superuser status for {}: {:?}", cfg.name, e));
return Err(e.into());
}
}
let psql = select_pg_path(&version).join(psql_binary_name());
logger.log("info", format!("Replaying cluster dump for {} via {:?}", cfg.name, psql));
let start = Instant::now();
let output = Command::new(&psql)
.arg("--host").arg(&cfg.host)
.arg("--port").arg(cfg.port.to_string())
.arg("--username").arg(&cfg.username)
.arg("--dbname").arg("postgres")
.arg("-f").arg(&restore_file)
.envs(env)
.output();
let duration_ms = start.elapsed().as_millis() as f64;
match output {
Ok(o) => {
let stderr = String::from_utf8_lossy(&o.stderr).to_string();
let stdout = String::from_utf8_lossy(&o.stdout).to_string();
let combined = format!("{}{}", stdout, stderr);
let exit_code = o.status.code().unwrap_or(-1);
if o.status.success() {
logger.log_command("psql", if combined.is_empty() { None } else { Some(combined) }, Some(0), Some(duration_ms));
logger.log("info", format!("Cluster restore completed for {}", cfg.name));
Ok(())
} else {
logger.log_command("psql", if combined.is_empty() { None } else { Some(combined) }, Some(exit_code), Some(duration_ms));
anyhow::bail!("Cluster restore (psql) failed for {}", cfg.name);
}
}
Err(e) => {
logger.log_command("psql", Some(e.to_string()), Some(-1), Some(duration_ms));
Err(e.into())
}
}
})
.await?
}
+27 -22
View File
@@ -33,32 +33,21 @@ pub async fn server_version(cfg: &DatabaseConfig) -> Result<String> {
Ok(version)
}
/// Resolves the `bin` directory of a PostgreSQL installation for the given
/// major version, in a cross-platform way.
///
/// Resolution order:
/// 1. The `PG_BIN_DIR` environment variable, if set, is used as-is. This
/// allows users/CI to override detection for non-standard installs
/// (e.g. portable PostgreSQL distributions, custom install locations).
/// 2. Platform-specific default install locations (Debian/Ubuntu packages,
/// the official Windows installer, Homebrew/Postgres.app on macOS, and
/// common RPM-based layouts on other Linux distros).
/// 3. A `PATH` lookup for `pg_dump` (`pg_dump.exe` on Windows), returning
/// its parent directory.
/// 4. The historical Debian/Ubuntu path as a last-resort fallback, so the
/// function keeps returning a `PathBuf` (never panics) even when nothing
/// was found, preserving the previous behavior for callers.
///
/// The override is sourced from `CONFIG.pg_bin_dir` (the `PG_BIN_DIR`
/// environment variable). An empty value means "unset" and falls through to
/// detection.
pub async fn is_superuser(cfg: &DatabaseConfig) -> Result<bool> {
let client = connect(cfg).await?;
let is_super: bool = client
.query_one("SELECT current_setting('is_superuser') = 'on';", &[])
.await?
.get(0);
Ok(is_super)
}
pub fn select_pg_path(version: &str) -> std::path::PathBuf {
select_pg_path_with(version, &CONFIG.pg_bin_dir)
}
/// Inner resolver behind [`select_pg_path`], parameterized over the
/// `PG_BIN_DIR` override. Kept pure (no env / no `CONFIG` access) so it is
/// unit-testable without mutating process-global state.
pub(crate) fn select_pg_path_with(version: &str, pg_bin_dir: &str) -> std::path::PathBuf {
let major = version.split('.').next().unwrap_or("17");
@@ -109,6 +98,22 @@ pub(crate) fn pg_dump_binary_name() -> &'static str {
}
}
pub(crate) fn pg_dumpall_binary_name() -> &'static str {
if cfg!(target_os = "windows") {
"pg_dumpall.exe"
} else {
"pg_dumpall"
}
}
pub(crate) fn psql_binary_name() -> &'static str {
if cfg!(target_os = "windows") {
"psql.exe"
} else {
"psql"
}
}
pub(crate) fn pg_dump_exists_in(dir: &std::path::Path) -> bool {
dir.join(pg_dump_binary_name()).is_file()
}
+1
View File
@@ -1,4 +1,5 @@
pub mod backup;
pub(crate) mod cluster;
pub(crate) mod connection;
pub mod database;
mod format;