validation: pruning for multiple chainstates

Introduces ChainstateManager::GetPruneRange().

The prune budget is split evenly between the number of chainstates,
however the prune budget may be exceeded if the resulting shares are
beneath `MIN_DISK_SPACE_FOR_BLOCK_FILES`.

Author: jamesob

doc/release-notes-27596.md

@@ -0,0 +1,7 @@
+Pruning
+-------
+
+When using assumeutxo with `-prune`, the prune budget may be exceeded if it is set
+lower than 1100MB (i.e. `MIN_DISK_SPACE_FOR_BLOCK_FILES * 2`). Prune budget is normally
+split evenly across each chainstate, unless the resulting prune budget per chainstate
+is beneath `MIN_DISK_SPACE_FOR_BLOCK_FILES` in which case that value will be used.

src/node/blockstorage.cpp

@@ -257,40 +257,56 @@ void BlockManager::PruneOneBlockFile(const int fileNumber)
     m_dirty_fileinfo.insert(fileNumber);
 }
 
-void BlockManager::FindFilesToPruneManual(std::set<int>& setFilesToPrune, int nManualPruneHeight, int chain_tip_height)
+void BlockManager::FindFilesToPruneManual(
+    std::set<int>& setFilesToPrune,
+    int nManualPruneHeight,
+    const Chainstate& chain,
+    ChainstateManager& chainman)
 {
     assert(IsPruneMode() && nManualPruneHeight > 0);
 
     LOCK2(cs_main, cs_LastBlockFile);
-    if (chain_tip_height < 0) {
+    if (chain.m_chain.Height() < 0) {
         return;
     }
 
-    // last block to prune is the lesser of (user-specified height, MIN_BLOCKS_TO_KEEP from the tip)
-    unsigned int nLastBlockWeCanPrune = std::min((unsigned)nManualPruneHeight, chain_tip_height - MIN_BLOCKS_TO_KEEP);
+    const auto [min_block_to_prune, last_block_can_prune] = chainman.GetPruneRange(chain, nManualPruneHeight);
+
     int count = 0;
     for (int fileNumber = 0; fileNumber < m_last_blockfile; fileNumber++) {
-        if (m_blockfile_info[fileNumber].nSize == 0 || m_blockfile_info[fileNumber].nHeightLast > nLastBlockWeCanPrune) {
+        const auto& fileinfo = m_blockfile_info[fileNumber];
+        if (fileinfo.nSize == 0 || fileinfo.nHeightLast > (unsigned)last_block_can_prune || fileinfo.nHeightFirst < (unsigned)min_block_to_prune) {
             continue;
         }
+
         PruneOneBlockFile(fileNumber);
         setFilesToPrune.insert(fileNumber);
         count++;
     }
-    LogPrintf("Prune (Manual): prune_height=%d removed %d blk/rev pairs\n", nLastBlockWeCanPrune, count);
+    LogPrintf("[%s] Prune (Manual): prune_height=%d removed %d blk/rev pairs\n",
+        chain.GetRole(), last_block_can_prune, count);
 }
 
-void BlockManager::FindFilesToPrune(std::set<int>& setFilesToPrune, uint64_t nPruneAfterHeight, int chain_tip_height, int prune_height, bool is_ibd)
+void BlockManager::FindFilesToPrune(
+    std::set<int>& setFilesToPrune,
+    int last_prune,
+    const Chainstate& chain,
+    ChainstateManager& chainman)
 {
     LOCK2(cs_main, cs_LastBlockFile);
-    if (chain_tip_height < 0 || GetPruneTarget() == 0) {
+    // Distribute our -prune budget over all chainstates.
+    const auto target = std::max(
+        MIN_DISK_SPACE_FOR_BLOCK_FILES, GetPruneTarget() / chainman.GetAll().size());
+
+    if (chain.m_chain.Height() < 0 || target == 0) {
         return;
     }
-    if ((uint64_t)chain_tip_height <= nPruneAfterHeight) {
+    if (static_cast<uint64_t>(chain.m_chain.Height()) <= chainman.GetParams().PruneAfterHeight()) {
         return;
     }
 
-    unsigned int nLastBlockWeCanPrune{(unsigned)std::min(prune_height, chain_tip_height - static_cast<int>(MIN_BLOCKS_TO_KEEP))};
+    const auto [min_block_to_prune, last_block_can_prune] = chainman.GetPruneRange(chain, last_prune);
+
     uint64_t nCurrentUsage = CalculateCurrentUsage();
     // We don't check to prune until after we've allocated new space for files
     // So we should leave a buffer under our target to account for another allocation
@@ -299,29 +315,31 @@ void BlockManager::FindFilesToPrune(std::set<int>& setFilesToPrune, uint64_t nPr
     uint64_t nBytesToPrune;
     int count = 0;
 
-    if (nCurrentUsage + nBuffer >= GetPruneTarget()) {
+    if (nCurrentUsage + nBuffer >= target) {
         // On a prune event, the chainstate DB is flushed.
         // To avoid excessive prune events negating the benefit of high dbcache
         // values, we should not prune too rapidly.
         // So when pruning in IBD, increase the buffer a bit to avoid a re-prune too soon.
-        if (is_ibd) {
+        if (chainman.IsInitialBlockDownload()) {
             // Since this is only relevant during IBD, we use a fixed 10%
-            nBuffer += GetPruneTarget() / 10;
+            nBuffer += target / 10;
         }
 
         for (int fileNumber = 0; fileNumber < m_last_blockfile; fileNumber++) {
-            nBytesToPrune = m_blockfile_info[fileNumber].nSize + m_blockfile_info[fileNumber].nUndoSize;
+            const auto& fileinfo = m_blockfile_info[fileNumber];
+            nBytesToPrune = fileinfo.nSize + fileinfo.nUndoSize;
 
-            if (m_blockfile_info[fileNumber].nSize == 0) {
+            if (fileinfo.nSize == 0) {
                 continue;
             }
 
-            if (nCurrentUsage + nBuffer < GetPruneTarget()) { // are we below our target?
+            if (nCurrentUsage + nBuffer < target) { // are we below our target?
                 break;
             }
 
-            // don't prune files that could have a block within MIN_BLOCKS_TO_KEEP of the main chain's tip but keep scanning
-            if (m_blockfile_info[fileNumber].nHeightLast > nLastBlockWeCanPrune) {
+            // don't prune files that could have a block that's not within the allowable
+            // prune range for the chain being pruned.
+            if (fileinfo.nHeightLast > (unsigned)last_block_can_prune || fileinfo.nHeightFirst < (unsigned)min_block_to_prune) {
                 continue;
             }
 
@@ -333,10 +351,10 @@ void BlockManager::FindFilesToPrune(std::set<int>& setFilesToPrune, uint64_t nPr
         }
     }
 
-    LogPrint(BCLog::PRUNE, "target=%dMiB actual=%dMiB diff=%dMiB max_prune_height=%d removed %d blk/rev pairs\n",
-             GetPruneTarget() / 1024 / 1024, nCurrentUsage / 1024 / 1024,
-             (int64_t(GetPruneTarget()) - int64_t(nCurrentUsage)) / 1024 / 1024,
-             nLastBlockWeCanPrune, count);
+    LogPrint(BCLog::PRUNE, "[%s] target=%dMiB actual=%dMiB diff=%dMiB min_height=%d max_prune_height=%d removed %d blk/rev pairs\n",
+             chain.GetRole(), target / 1024 / 1024, nCurrentUsage / 1024 / 1024,
+             (int64_t(target) - int64_t(nCurrentUsage)) / 1024 / 1024,
+             min_block_to_prune, last_block_can_prune, count);
 }
 
 void BlockManager::UpdatePruneLock(const std::string& name, const PruneLockInfo& lock_info) {

src/node/blockstorage.h

@@ -36,6 +36,7 @@ class CBlockUndo;
 class CChainParams;
 class Chainstate;
 class ChainstateManager;
+enum class ChainstateRole;
 struct CCheckpointData;
 struct FlatFilePos;
 namespace Consensus {
@@ -138,7 +139,11 @@ class BlockManager
     bool UndoWriteToDisk(const CBlockUndo& blockundo, FlatFilePos& pos, const uint256& hashBlock) const;
 
     /* Calculate the block/rev files to delete based on height specified by user with RPC command pruneblockchain */
-    void FindFilesToPruneManual(std::set<int>& setFilesToPrune, int nManualPruneHeight, int chain_tip_height);
+    void FindFilesToPruneManual(
+        std::set<int>& setFilesToPrune,
+        int nManualPruneHeight,
+        const Chainstate& chain,
+        ChainstateManager& chainman);
 
     /**
      * Prune block and undo files (blk???.dat and rev???.dat) so that the disk space used is less than a user-defined target.
@@ -154,8 +159,13 @@ class BlockManager
      * A db flag records the fact that at least some block files have been pruned.
      *
      * @param[out]   setFilesToPrune   The set of file indices that can be unlinked will be returned
+     * @param        last_prune        The last height we're able to prune, according to the prune locks
      */
-    void FindFilesToPrune(std::set<int>& setFilesToPrune, uint64_t nPruneAfterHeight, int chain_tip_height, int prune_height, bool is_ibd);
+    void FindFilesToPrune(
+        std::set<int>& setFilesToPrune,
+        int last_prune,
+        const Chainstate& chain,
+        ChainstateManager& chainman);
 
     RecursiveMutex cs_LastBlockFile;
     std::vector<CBlockFileInfo> m_blockfile_info;

src/validation.cpp

@@ -69,6 +69,7 @@
 #include <optional>
 #include <string>
 #include <utility>
+#include <tuple>
 
 using kernel::CCoinsStats;
 using kernel::CoinStatsHashType;
@@ -2552,11 +2553,14 @@ bool Chainstate::FlushStateToDisk(
             if (nManualPruneHeight > 0) {
                 LOG_TIME_MILLIS_WITH_CATEGORY("find files to prune (manual)", BCLog::BENCH);
 
-                m_blockman.FindFilesToPruneManual(setFilesToPrune, std::min(last_prune, nManualPruneHeight), m_chain.Height());
+                m_blockman.FindFilesToPruneManual(
+                    setFilesToPrune,
+                    std::min(last_prune, nManualPruneHeight),
+                    *this, m_chainman);
             } else {
                 LOG_TIME_MILLIS_WITH_CATEGORY("find files to prune", BCLog::BENCH);
 
-                m_blockman.FindFilesToPrune(setFilesToPrune, m_chainman.GetParams().PruneAfterHeight(), m_chain.Height(), last_prune, m_chainman.IsInitialBlockDownload());
+                m_blockman.FindFilesToPrune(setFilesToPrune, last_prune, *this, m_chainman);
                 m_blockman.m_check_for_pruning = false;
             }
             if (!setFilesToPrune.empty()) {
@@ -5939,3 +5943,30 @@ Chainstate& ChainstateManager::GetChainstateForIndexing()
     // indexed.
     return (this->GetAll().size() > 1) ? *m_ibd_chainstate : *m_active_chainstate;
 }
+
+std::pair<int, int> ChainstateManager::GetPruneRange(const Chainstate& chainstate, int last_height_can_prune)
+{
+    if (chainstate.m_chain.Height() <= 0) {
+        return {0, 0};
+    }
+    int prune_start{0};
+
+    if (this->GetAll().size() > 1 && m_snapshot_chainstate.get() == &chainstate) {
+        // Leave the blocks in the background IBD chain alone if we're pruning
+        // the snapshot chain.
+        prune_start = *Assert(GetSnapshotBaseHeight()) + 1;
+    }
+
+    int max_prune = std::max<int>(
+        0, chainstate.m_chain.Height() - static_cast<int>(MIN_BLOCKS_TO_KEEP));
+
+    // last block to prune is the lesser of (caller-specified height, MIN_BLOCKS_TO_KEEP from the tip)
+    //
+    // While you might be tempted to prune the background chainstate more
+    // aggressively (i.e. fewer MIN_BLOCKS_TO_KEEP), this won't work with index
+    // building - specifically blockfilterindex requires undo data, and if
+    // we don't maintain this trailing window, we hit indexing failures.
+    int prune_end = std::min(last_height_can_prune, max_prune);
+
+    return {prune_start, prune_end};
+}

src/validation.h

@@ -885,10 +885,6 @@ class ChainstateManager
     /** Most recent headers presync progress update, for rate-limiting. */
     std::chrono::time_point<std::chrono::steady_clock> m_last_presync_update GUARDED_BY(::cs_main) {};
 
-    //! Return the height of the base block of the snapshot in use, if one exists, else
-    //! nullopt.
-    std::optional<int> GetSnapshotBaseHeight() const EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
-
     std::array<ThresholdConditionCache, VERSIONBITS_NUM_BITS> m_warningcache GUARDED_BY(::cs_main);
 
     //! Return true if a chainstate is considered usable.
@@ -1241,6 +1237,16 @@ class ChainstateManager
     //!   fully validated chain.
     Chainstate& GetChainstateForIndexing() EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
 
+    //! Return the [start, end] (inclusive) of block heights we can prune.
+    //!
+    //! start > end is possible, meaning no blocks can be pruned.
+    std::pair<int, int> GetPruneRange(
+        const Chainstate& chainstate, int last_height_can_prune) EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
+
+    //! Return the height of the base block of the snapshot in use, if one exists, else
+    //! nullopt.
+    std::optional<int> GetSnapshotBaseHeight() const EXCLUSIVE_LOCKS_REQUIRED(::cs_main);
+
     ~ChainstateManager();
 };