all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: Dominik Csapak <d.csapak@proxmox.com>
To: pbs-devel@lists.proxmox.com
Subject: [pbs-devel] [PATCH proxmox-backup 2/3] verify: keep also track of corrupt chunks
Date: Thu, 30 Jul 2020 09:09:04 +0200	[thread overview]
Message-ID: <20200730070905.10735-2-d.csapak@proxmox.com> (raw)
In-Reply-To: <20200730070905.10735-1-d.csapak@proxmox.com>

so that we do not have to verify a corrupt one multiple times

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
---
 src/api2/admin/datastore.rs |  3 ++-
 src/backup/verify.rs        | 46 +++++++++++++++++++++++++++++--------
 2 files changed, 39 insertions(+), 10 deletions(-)

diff --git a/src/api2/admin/datastore.rs b/src/api2/admin/datastore.rs
index c720231c..5e6d5720 100644
--- a/src/api2/admin/datastore.rs
+++ b/src/api2/admin/datastore.rs
@@ -476,7 +476,8 @@ pub fn verify(
         {
             let success = if let Some(backup_dir) = backup_dir {
                 let mut verified_chunks = HashSet::with_capacity(1024*16);
-                verify_backup_dir(&datastore, &backup_dir, &mut verified_chunks, &worker)?
+                let mut corrupt_chunks = HashSet::with_capacity(64);
+                verify_backup_dir(&datastore, &backup_dir, &mut verified_chunks, &mut corrupt_chunks, &worker)?
             } else if let Some(backup_group) = backup_group {
                 verify_backup_group(&datastore, &backup_group, &worker)?
             } else {
diff --git a/src/backup/verify.rs b/src/backup/verify.rs
index 9e309b12..58b91bc9 100644
--- a/src/backup/verify.rs
+++ b/src/backup/verify.rs
@@ -39,6 +39,7 @@ fn verify_index_chunks(
     datastore: &DataStore,
     index: Box<dyn IndexFile>,
     verified_chunks: &mut HashSet<[u8;32]>,
+    corrupt_chunks: &mut HashSet<[u8; 32]>,
     worker: &WorkerTask,
 ) -> Result<(), Error> {
 
@@ -51,11 +52,18 @@ fn verify_index_chunks(
         let size = info.range.end - info.range.start;
 
         if !verified_chunks.contains(&info.digest) {
-            if let Err(err) = datastore.verify_stored_chunk(&info.digest, size) {
-                worker.log(format!("{}", err));
-                errors += 1;
+            if !corrupt_chunks.contains(&info.digest) {
+                if let Err(err) = datastore.verify_stored_chunk(&info.digest, size) {
+                    corrupt_chunks.insert(info.digest);
+                    worker.log(format!("{}", err));
+                    errors += 1;
+                } else {
+                    verified_chunks.insert(info.digest);
+                }
             } else {
-                verified_chunks.insert(info.digest);
+                let digest_str = proxmox::tools::digest_to_hex(&info.digest);
+                worker.log(format!("chunk {} was marked as corrupt", digest_str));
+                errors += 1;
             }
         }
     }
@@ -72,6 +80,7 @@ fn verify_fixed_index(
     backup_dir: &BackupDir,
     info: &FileInfo,
     verified_chunks: &mut HashSet<[u8;32]>,
+    corrupt_chunks: &mut HashSet<[u8;32]>,
     worker: &WorkerTask,
 ) -> Result<(), Error> {
 
@@ -89,7 +98,7 @@ fn verify_fixed_index(
         bail!("wrong index checksum");
     }
 
-    verify_index_chunks(datastore, Box::new(index), verified_chunks, worker)
+    verify_index_chunks(datastore, Box::new(index), verified_chunks, corrupt_chunks, worker)
 }
 
 fn verify_dynamic_index(
@@ -97,6 +106,7 @@ fn verify_dynamic_index(
     backup_dir: &BackupDir,
     info: &FileInfo,
     verified_chunks: &mut HashSet<[u8;32]>,
+    corrupt_chunks: &mut HashSet<[u8;32]>,
     worker: &WorkerTask,
 ) -> Result<(), Error> {
 
@@ -114,7 +124,7 @@ fn verify_dynamic_index(
         bail!("wrong index checksum");
     }
 
-    verify_index_chunks(datastore, Box::new(index), verified_chunks, worker)
+    verify_index_chunks(datastore, Box::new(index), verified_chunks, corrupt_chunks, worker)
 }
 
 /// Verify a single backup snapshot
@@ -130,6 +140,7 @@ pub fn verify_backup_dir(
     datastore: &DataStore,
     backup_dir: &BackupDir,
     verified_chunks: &mut HashSet<[u8;32]>,
+    corrupt_chunks: &mut HashSet<[u8;32]>,
     worker: &WorkerTask
 ) -> Result<bool, Error> {
 
@@ -149,8 +160,24 @@ pub fn verify_backup_dir(
         let result = proxmox::try_block!({
             worker.log(format!("  check {}", info.filename));
             match archive_type(&info.filename)? {
-                ArchiveType::FixedIndex => verify_fixed_index(&datastore, &backup_dir, info, verified_chunks, worker),
-                ArchiveType::DynamicIndex => verify_dynamic_index(&datastore, &backup_dir, info, verified_chunks, worker),
+                ArchiveType::FixedIndex =>
+                    verify_fixed_index(
+                        &datastore,
+                        &backup_dir,
+                        info,
+                        verified_chunks,
+                        corrupt_chunks,
+                        worker
+                    ),
+                ArchiveType::DynamicIndex =>
+                    verify_dynamic_index(
+                        &datastore,
+                        &backup_dir,
+                        info,
+                        verified_chunks,
+                        corrupt_chunks,
+                        worker
+                    ),
                 ArchiveType::Blob => verify_blob(&datastore, &backup_dir, info),
             }
         });
@@ -189,10 +216,11 @@ pub fn verify_backup_group(datastore: &DataStore, group: &BackupGroup, worker: &
     let mut error_count = 0;
 
     let mut verified_chunks = HashSet::with_capacity(1024*16); // start with 16384 chunks (up to 65GB)
+    let mut corrupt_chunks = HashSet::with_capacity(64); // start with 64 chunks since we assume there are few corrupt ones
 
     BackupInfo::sort_list(&mut list, false); // newest first
     for info in list {
-        if !verify_backup_dir(datastore, &info.backup_dir, &mut verified_chunks, worker)? {
+        if !verify_backup_dir(datastore, &info.backup_dir, &mut verified_chunks, &mut corrupt_chunks, worker)?{
             error_count += 1;
         }
     }
-- 
2.20.1





  reply	other threads:[~2020-07-30  7:09 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-30  7:09 [pbs-devel] [PATCH proxmox-backup 1/3] verify: check all chunks of an index, even if we encounter a corrupt one Dominik Csapak
2020-07-30  7:09 ` Dominik Csapak [this message]
2020-07-30  7:09 ` [pbs-devel] [PATCH proxmox-backup 3/3] verify: keep track and log which dirs failed the verification Dominik Csapak
2020-07-30  7:40 ` [pbs-devel] applied: [PATCH proxmox-backup 1/3] verify: check all chunks of an index, even if we encounter a corrupt one Dietmar Maurer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200730070905.10735-2-d.csapak@proxmox.com \
    --to=d.csapak@proxmox.com \
    --cc=pbs-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal