From d3d566f7bdb6198f87e3db051aa407f626fa183c Mon Sep 17 00:00:00 2001 From: Aaron Lauterer Date: Tue, 21 Jul 2020 17:29:40 +0200 Subject: [PATCH] GC: use time pre phase1 to calculate min_atime in phase2 Used chunks are marked in phase1 of the garbage collection process by using the atime property. Each used chunk gets touched so that the atime gets updated (if older than 24h, see relatime). Should there ever be a situation in which the phase1 in the GC run needs a very long time to finish, it could happen that the grace period calculated in phase2 is not long enough and thus the marking of the chunks (atime) becomes invalid. This would result in the removal of needed chunks. Even though the likelyhood of this happening is very low, using the timestamp from right before phase1 is started, to calculate the grace period in phase2 should avoid this situation. Signed-off-by: Aaron Lauterer --- src/backup/chunk_store.rs | 5 ++--- src/backup/datastore.rs | 2 +- 2 files changed, 3 insertions(+), 4 deletions(-) diff --git a/src/backup/chunk_store.rs b/src/backup/chunk_store.rs index 5cc944c9..360745d2 100644 --- a/src/backup/chunk_store.rs +++ b/src/backup/chunk_store.rs @@ -275,14 +275,13 @@ impl ChunkStore { pub fn sweep_unused_chunks( &self, oldest_writer: i64, + phase1_start_time: i64, status: &mut GarbageCollectionStatus, worker: &WorkerTask, ) -> Result<(), Error> { use nix::sys::stat::fstatat; - let now = unsafe { libc::time(std::ptr::null_mut()) }; - - let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime) + let mut min_atime = phase1_start_time - 3600*24; // at least 24h (see mount option relatime) if oldest_writer < min_atime { min_atime = oldest_writer; diff --git a/src/backup/datastore.rs b/src/backup/datastore.rs index a8bb282b..ffd64b81 100644 --- a/src/backup/datastore.rs +++ b/src/backup/datastore.rs @@ -492,7 +492,7 @@ impl DataStore { self.mark_used_chunks(&mut gc_status, &worker)?; worker.log("Start GC phase2 (sweep unused chunks)"); - self.chunk_store.sweep_unused_chunks(oldest_writer, &mut gc_status, &worker)?; + self.chunk_store.sweep_unused_chunks(oldest_writer, now, &mut gc_status, &worker)?; worker.log(&format!("Removed bytes: {}", gc_status.removed_bytes)); worker.log(&format!("Removed chunks: {}", gc_status.removed_chunks));