From: Christian Ebner <c.ebner@proxmox.com>
To: pbs-devel@lists.proxmox.com
Subject: [pbs-devel] [PATCH proxmox-backup v6 15/21] GC: guard missing marker file insertion for s3 backed stores
Date: Fri, 14 Nov 2025 14:18:55 +0100 [thread overview]
Message-ID: <20251114131901.441650-16-c.ebner@proxmox.com> (raw)
In-Reply-To: <20251114131901.441650-1-c.ebner@proxmox.com>
The chunk marker file should only ever be missing if the local
datastore cache has been recreated (e.g. after setup on a new PBS
instance while reusing the s3 bucket contents) or by manual user
interaction. Garbage collection does re-create the marker in these
cases.
Guard this marker file creation by the per-chunk file lock to not run
into races with chunk insertion operations. Since this requires to
stat the chunk marker file and locking is expensive, first check if
the marker exists without holding a lock, only then lock and retry.
Since the first chunk file existence check is already performed
anyways move the logic to be within the non-existing branch thereof.
By making this happen after touching potential bad chunks, this will
allow to check these beforehand.
Signed-off-by: Christian Ebner <c.ebner@proxmox.com>
---
pbs-datastore/src/datastore.rs | 46 ++++++++++++++++++----------------
1 file changed, 25 insertions(+), 21 deletions(-)
diff --git a/pbs-datastore/src/datastore.rs b/pbs-datastore/src/datastore.rs
index d71106ea8..9c0ce9859 100644
--- a/pbs-datastore/src/datastore.rs
+++ b/pbs-datastore/src/datastore.rs
@@ -1323,36 +1323,40 @@ impl DataStore {
}
if !self.inner.chunk_store.cond_touch_chunk(digest, false)? {
- let hex = hex::encode(digest);
- warn!(
- "warning: unable to access non-existent chunk {hex}, required by {file_name:?}"
- );
-
+ let (chunk_path, _digest_str) = self.chunk_path(digest);
// touch any corresponding .bad files to keep them around, meaning if a chunk is
// rewritten correctly they will be removed automatically, as well as if no index
// file requires the chunk anymore (won't get to this loop then)
for i in 0..=9 {
let bad_ext = format!("{i}.bad");
- let mut bad_path = PathBuf::new();
- bad_path.push(self.chunk_path(digest).0);
+ let mut bad_path = chunk_path.clone();
bad_path.set_extension(bad_ext);
self.inner.chunk_store.cond_touch_path(&bad_path, false)?;
}
- }
- if let Some(ref _s3_client) = s3_client {
- // Update atime on local cache marker files.
- if !self.inner.chunk_store.cond_touch_chunk(digest, false)? {
- let (chunk_path, _digest) = self.chunk_path(digest);
- // Insert empty file as marker to tell GC phase2 that this is
- // a chunk still in-use, so to keep in the S3 object store.
- std::fs::File::options()
- .write(true)
- .create_new(true)
- .open(&chunk_path)
- .with_context(|| {
- format!("failed to create marker for chunk {}", hex::encode(digest))
- })?;
+ if let Some(ref _s3_client) = s3_client {
+ // Do not retry here, this is very unlikely to happen as chunk markers will
+ // most likely only be missing if the local cache store was recreated.
+ let _guard = self
+ .inner
+ .chunk_store
+ .lock_chunk(digest, CHUNK_LOCK_TIMEOUT)?;
+ if !self.inner.chunk_store.cond_touch_chunk(digest, false)? {
+ // Insert empty file as marker to tell GC phase2 that this is
+ // a chunk still in-use, so to keep in the S3 object store.
+ std::fs::File::options()
+ .write(true)
+ .create_new(true)
+ .open(&chunk_path)
+ .with_context(|| {
+ format!("failed to create marker for chunk {}", hex::encode(digest))
+ })?;
+ }
+ } else {
+ let hex = hex::encode(digest);
+ warn!(
+ "warning: unable to access non-existent chunk {hex}, required by {file_name:?}"
+ );
}
}
}
--
2.47.3
_______________________________________________
pbs-devel mailing list
pbs-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel
next prev parent reply other threads:[~2025-11-14 13:18 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-11-14 13:18 [pbs-devel] [PATCH proxmox-backup v6 00/21] fix chunk upload/insert, rename corrupt chunks and GC race conditions for s3 backend Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 01/21] datastore: GC: drop overly verbose info message during s3 chunk sweep Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 02/21] chunk store: implement per-chunk file locking helper for s3 backend Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 03/21] datastore: acquire chunk store mutex lock when renaming corrupt chunk Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 04/21] datastore: get per-chunk file lock for chunk rename on s3 backend Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 05/21] fix #6961: datastore: verify: evict corrupt chunks from in-memory LRU cache Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 06/21] datastore: add locking to protect against races on chunk insert for s3 Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 07/21] GC: fix race with chunk upload/insert on s3 backends Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 08/21] chunk store: reduce exposure of clear_chunk() to crate only Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 09/21] chunk store: make chunk removal a helper method of the chunk store Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 10/21] store: split insert_chunk into wrapper + unsafe locked implementation Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 11/21] store: cache: move Mutex acquire to cache insertion Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 12/21] chunk store: rename cache-specific helpers Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 13/21] GC: cleanup chunk markers from cache in phase 3 on s3 backends Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 14/21] GC: touch bad chunk files independent of backend type Christian Ebner
2025-11-14 13:18 ` Christian Ebner [this message]
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 16/21] GC: s3: track if a chunk marker file is missing since a bad chunk Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 17/21] chunk store: add helpers marking missing local chunk markers as expected Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 18/21] GC: assure chunk exists on s3 store when creating missing chunk marker Christian Ebner
2025-11-14 13:18 ` [pbs-devel] [PATCH proxmox-backup v6 19/21] datastore: document s3 backend specific locking restrictions Christian Ebner
2025-11-14 13:19 ` [pbs-devel] [PATCH proxmox-backup v6 20/21] GC: fix: don't drop bad extension for S3 object to chunk path helper Christian Ebner
2025-11-14 13:19 ` [pbs-devel] [PATCH proxmox-backup v6 21/21] GC: clean up bad chunks from the filesystem only Christian Ebner
2025-11-14 13:34 ` [pbs-devel] [PATCH proxmox-backup v6 00/21] fix chunk upload/insert, rename corrupt chunks and GC race conditions for s3 backend Fabian Grünbichler
2025-11-14 22:14 ` Thomas Lamprecht
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=20251114131901.441650-16-c.ebner@proxmox.com \
--to=c.ebner@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox