From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits)) (No client certificate requested) by lists.proxmox.com (Postfix) with ESMTPS id 05D44608A0 for ; Thu, 15 Oct 2020 12:49:58 +0200 (CEST) Received: from firstgate.proxmox.com (localhost [127.0.0.1]) by firstgate.proxmox.com (Proxmox) with ESMTP id EA9C311FD1 for ; Thu, 15 Oct 2020 12:49:27 +0200 (CEST) Received: from proxmox-new.maurer-it.com (proxmox-new.maurer-it.com [212.186.127.180]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits)) (No client certificate requested) by firstgate.proxmox.com (Proxmox) with ESMTPS id ED6DF11FA9 for ; Thu, 15 Oct 2020 12:49:25 +0200 (CEST) Received: from proxmox-new.maurer-it.com (localhost.localdomain [127.0.0.1]) by proxmox-new.maurer-it.com (Proxmox) with ESMTP id B1B4D45D95 for ; Thu, 15 Oct 2020 12:49:25 +0200 (CEST) From: Stefan Reiter To: pbs-devel@lists.proxmox.com Date: Thu, 15 Oct 2020 12:49:14 +0200 Message-Id: <20201015104916.21170-3-s.reiter@proxmox.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20201015104916.21170-1-s.reiter@proxmox.com> References: <20201015104916.21170-1-s.reiter@proxmox.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-SPAM-LEVEL: Spam detection results: 0 AWL -0.036 Adjusted score from AWL reputation of From: address KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment RCVD_IN_DNSWL_MED -2.3 Sender listed at https://www.dnswl.org/, medium trust SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record URIBL_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to URIBL was blocked. See http://wiki.apache.org/spamassassin/DnsBlocklists#dnsbl-block for more information. [datastore.rs, manifest.rs, verify.rs, environment.rs] Subject: [pbs-devel] [PATCH v2 proxmox-backup 2/4] datastore: add manifest locking X-BeenThere: pbs-devel@lists.proxmox.com X-Mailman-Version: 2.1.29 Precedence: list List-Id: Proxmox Backup Server development discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 15 Oct 2020 10:49:58 -0000 Avoid races when updating manifest data by flocking a lock file. update_manifest is used to ensure updates always happen with the lock held. Snapshot deletion also acquires the lock, so it cannot interfere with an outstanding manifest write. Signed-off-by: Stefan Reiter --- v2: * Use seperate manifest lock file * Change store_manifest to update_manifest to force consumers to use correct locking * Don't hold lock across verify, reload manifest at the end * Update comments src/api2/admin/datastore.rs | 8 +++---- src/api2/backup/environment.rs | 13 ++++------ src/backup/datastore.rs | 43 +++++++++++++++++++++++++++++----- src/backup/manifest.rs | 1 + src/backup/verify.rs | 9 +++---- 5 files changed, 50 insertions(+), 24 deletions(-) diff --git a/src/api2/admin/datastore.rs b/src/api2/admin/datastore.rs index de39cd1b..4f15c1cd 100644 --- a/src/api2/admin/datastore.rs +++ b/src/api2/admin/datastore.rs @@ -1481,11 +1481,9 @@ fn set_notes( let allowed = (user_privs & PRIV_DATASTORE_READ) != 0; if !allowed { check_backup_owner(&datastore, backup_dir.group(), &userid)?; } - let (mut manifest, _) = datastore.load_manifest(&backup_dir)?; - - manifest.unprotected["notes"] = notes.into(); - - datastore.store_manifest(&backup_dir, manifest)?; + datastore.update_manifest(&backup_dir,|manifest| { + manifest.unprotected["notes"] = notes.into(); + }).map_err(|err| format_err!("unable to update manifest blob - {}", err))?; Ok(()) } diff --git a/src/api2/backup/environment.rs b/src/api2/backup/environment.rs index f00c2cd3..c4f166df 100644 --- a/src/api2/backup/environment.rs +++ b/src/api2/backup/environment.rs @@ -472,16 +472,11 @@ impl BackupEnvironment { bail!("backup does not contain valid files (file count == 0)"); } - // check manifest - let (mut manifest, _) = self.datastore.load_manifest(&self.backup_dir) - .map_err(|err| format_err!("unable to load manifest blob - {}", err))?; - + // check for valid manifest and store stats let stats = serde_json::to_value(state.backup_stat)?; - - manifest.unprotected["chunk_upload_stats"] = stats; - - self.datastore.store_manifest(&self.backup_dir, manifest) - .map_err(|err| format_err!("unable to store manifest blob - {}", err))?; + self.datastore.update_manifest(&self.backup_dir, |manifest| { + manifest.unprotected["chunk_upload_stats"] = stats; + }).map_err(|err| format_err!("unable to update manifest blob - {}", err))?; if let Some(base) = &self.last_backup { let path = self.datastore.snapshot_path(&base.backup_dir); diff --git a/src/backup/datastore.rs b/src/backup/datastore.rs index ca8ca438..7b2bee7e 100644 --- a/src/backup/datastore.rs +++ b/src/backup/datastore.rs @@ -3,17 +3,19 @@ use std::io::{self, Write}; use std::path::{Path, PathBuf}; use std::sync::{Arc, Mutex}; use std::convert::TryFrom; +use std::time::Duration; +use std::fs::File; use anyhow::{bail, format_err, Error}; use lazy_static::lazy_static; -use proxmox::tools::fs::{replace_file, CreateOptions}; +use proxmox::tools::fs::{replace_file, CreateOptions, open_file_locked}; use super::backup_info::{BackupGroup, BackupDir}; use super::chunk_store::ChunkStore; use super::dynamic_index::{DynamicIndexReader, DynamicIndexWriter}; use super::fixed_index::{FixedIndexReader, FixedIndexWriter}; -use super::manifest::{MANIFEST_BLOB_NAME, CLIENT_LOG_BLOB_NAME, BackupManifest}; +use super::manifest::{MANIFEST_BLOB_NAME, MANIFEST_LOCK_NAME, CLIENT_LOG_BLOB_NAME, BackupManifest}; use super::index::*; use super::{DataBlob, ArchiveType, archive_type}; use crate::config::datastore; @@ -235,9 +237,10 @@ impl DataStore { let full_path = self.snapshot_path(backup_dir); - let _guard; + let (_guard, _manifest_guard); if !force { _guard = lock_dir_noblock(&full_path, "snapshot", "possibly running or in use")?; + _manifest_guard = self.lock_manifest(backup_dir); } // Acquire lock and keep it during remove operation, so there's no @@ -665,8 +668,27 @@ impl DataStore { digest_str, err, )) - } + } + fn lock_manifest( + &self, + backup_dir: &BackupDir, + ) -> Result { + let mut path = self.base_path(); + path.push(backup_dir.relative_path()); + path.push(&MANIFEST_LOCK_NAME); + + // update_manifest should never take a long time, so if someone else has + // the lock we can simply block a bit and should get it soon + open_file_locked(&path, Duration::from_secs(5), true) + .map_err(|err| { + format_err!( + "unable to acquire manifest lock {:?} - {}", &path, err + ) + }) + } + + /// Load the manifest without a lock. Must not be written back. pub fn load_manifest( &self, backup_dir: &BackupDir, @@ -677,11 +699,19 @@ impl DataStore { Ok((manifest, raw_size)) } - pub fn store_manifest( + /// Update the manifest of the specified snapshot. Never write a manifest directly, + /// only use this method - anything else may break locking guarantees. + pub fn update_manifest( &self, backup_dir: &BackupDir, - manifest: BackupManifest, + update_fn: impl FnOnce(&mut BackupManifest), ) -> Result<(), Error> { + + let _guard = self.lock_manifest(backup_dir)?; + let (mut manifest, _) = self.load_manifest(&backup_dir)?; + + update_fn(&mut manifest); + let manifest = serde_json::to_value(manifest)?; let manifest = serde_json::to_string_pretty(&manifest)?; let blob = DataBlob::encode(manifest.as_bytes(), None, true)?; @@ -691,6 +721,7 @@ impl DataStore { path.push(backup_dir.relative_path()); path.push(MANIFEST_BLOB_NAME); + // atomic replace invalidates flock - no other writes past this point! replace_file(&path, raw_data, CreateOptions::new())?; Ok(()) diff --git a/src/backup/manifest.rs b/src/backup/manifest.rs index 609cc998..51980a07 100644 --- a/src/backup/manifest.rs +++ b/src/backup/manifest.rs @@ -8,6 +8,7 @@ use ::serde::{Deserialize, Serialize}; use crate::backup::{BackupDir, CryptMode, CryptConfig}; pub const MANIFEST_BLOB_NAME: &str = "index.json.blob"; +pub const MANIFEST_LOCK_NAME: &str = ".index.json.lck"; pub const CLIENT_LOG_BLOB_NAME: &str = "client.log.blob"; mod hex_csum { diff --git a/src/backup/verify.rs b/src/backup/verify.rs index 05b6ba86..ea3fa760 100644 --- a/src/backup/verify.rs +++ b/src/backup/verify.rs @@ -300,7 +300,7 @@ pub fn verify_backup_dir( return Ok(true); } - let mut manifest = match datastore.load_manifest(&backup_dir) { + let manifest = match datastore.load_manifest(&backup_dir) { Ok((manifest, _)) => manifest, Err(err) => { task_log!( @@ -367,9 +367,10 @@ pub fn verify_backup_dir( state: verify_result, upid, }; - manifest.unprotected["verify_state"] = serde_json::to_value(verify_state)?; - datastore.store_manifest(&backup_dir, manifest) - .map_err(|err| format_err!("unable to store manifest blob - {}", err))?; + let verify_state = serde_json::to_value(verify_state)?; + datastore.update_manifest(&backup_dir, |manifest| { + manifest.unprotected["verify_state"] = verify_state; + }).map_err(|err| format_err!("unable to update manifest blob - {}", err))?; Ok(error_count == 0) } -- 2.20.1