add separate PreCache step for GcReferenceChecker (#578)

- Improvement: GCv2: Use separate PreCache step to improve concurrency when checking references
- Improvement: GCv2: Improved verbose logging
- Improvement: GCv2: Sort chunks to read by block/offset when finding references
- Improvement: GCv2: Exit as soon as no more unreferenced items are left

1 files changed, 29 insertions, 14 deletions

diff --git a/src/zenserver/projectstore/projectstore.cpp b/src/zenserver/projectstore/projectstore.cpp
index dde824b6f..c6097dea2 100644
--- a/src/zenserver/projectstore/projectstore.cpp
+++ b/src/zenserver/projectstore/projectstore.cpp
@@ -3311,11 +3311,15 @@ ProjectStore::RemoveExpiredData(GcCtx& Ctx, GcStats& Stats)
 class ProjectStoreReferenceChecker : public GcReferenceChecker
 {
 public:
-	ProjectStoreReferenceChecker(GcCtx& Ctx, ProjectStore::Oplog& Owner, bool PreCache) : m_Oplog(Owner)
+	ProjectStoreReferenceChecker(ProjectStore::Oplog& Owner, bool PreCache) : m_Oplog(Owner), m_PreCache(PreCache) {}
+
+	virtual ~ProjectStoreReferenceChecker() {}
+
+	virtual void PreCache(GcCtx& Ctx) override
 	{
-		if (PreCache)
+		if (m_PreCache)
 		{
-			ZEN_TRACE_CPU("Store::ReferencesPreCache");
+			ZEN_TRACE_CPU("Store::PreCache");
 
 			Stopwatch  Timer;
 			const auto _ = MakeGuard([&] {
@@ -3323,7 +3327,7 @@ public:
 				{
 					return;
 				}
-				ZEN_INFO("GCV2: projectstore [LOCKSTATE] '{}': precached {} references in {} from {}/{}",
+				ZEN_INFO("GCV2: projectstore [PRECACHE] '{}': precached {} references in {} from {}/{}",
 						 m_Oplog.m_BasePath,
 						 m_References.size(),
 						 NiceTimeSpanMs(Timer.GetElapsedTimeMs()),
@@ -3332,14 +3336,17 @@ public:
 			});
 
 			RwLock::SharedLockScope __(m_Oplog.m_OplogLock);
-			m_Oplog.IterateOplog(
-				[&](CbObjectView Op) { Op.IterateAttachments([&](CbFieldView Visitor) { m_References.insert(Visitor.AsAttachment()); }); });
+			if (Ctx.IsCancelledFlag)
+			{
+				return;
+			}
+			m_Oplog.IterateOplog([&](CbObjectView Op) {
+				Op.IterateAttachments([&](CbFieldView Visitor) { m_References.emplace_back(Visitor.AsAttachment()); });
+			});
 			m_PreCachedLsn = m_Oplog.GetMaxOpIndex();
 		}
 	}
 
-	virtual ~ProjectStoreReferenceChecker() {}
-
 	virtual void LockState(GcCtx& Ctx) override
 	{
 		ZEN_TRACE_CPU("Store::LockState");
@@ -3363,9 +3370,10 @@ public:
 		{
 			// TODO: Maybe we could just check the added oplog entries - we might get a few extra references from obsolete entries
 			// but I don't think that would be critical
-			m_References.clear();
-			m_Oplog.IterateOplog(
-				[&](CbObjectView Op) { Op.IterateAttachments([&](CbFieldView Visitor) { m_References.insert(Visitor.AsAttachment()); }); });
+			m_References.resize(0);
+			m_Oplog.IterateOplog([&](CbObjectView Op) {
+				Op.IterateAttachments([&](CbFieldView Visitor) { m_References.emplace_back(Visitor.AsAttachment()); });
+			});
 		}
 	}
 
@@ -3391,12 +3399,19 @@ public:
 
 		for (const IoHash& ReferenceHash : m_References)
 		{
-			IoCids.erase(ReferenceHash);
+			if (IoCids.erase(ReferenceHash) == 1)
+			{
+				if (IoCids.empty())
+				{
+					return;
+				}
+			}
 		}
 	}
 	ProjectStore::Oplog&					 m_Oplog;
+	bool									 m_PreCache;
 	std::unique_ptr<RwLock::SharedLockScope> m_OplogLock;
-	HashSet									 m_References;
+	std::vector<IoHash>						 m_References;
 	int										 m_PreCachedLsn = -1;
 };
 
@@ -3446,7 +3461,7 @@ ProjectStore::CreateReferenceCheckers(GcCtx& Ctx)
 				ProjectStore::Oplog* Oplog		 = Project->OpenOplog(OpLogId);
 				GcClock::TimePoint	 Now		 = GcClock::Now();
 				bool				 TryPreCache = Project->LastOplogAccessTime(OpLogId) < (Now - std::chrono::minutes(5));
-				Checkers.emplace_back(new ProjectStoreReferenceChecker(Ctx, *Oplog, TryPreCache));
+				Checkers.emplace_back(new ProjectStoreReferenceChecker(*Oplog, TryPreCache));
 			}
 			OplogCount += OpLogs.size();
 		}