Track missing vertex lookups in metrics (#3277)

When looking up a VertexID, the entry might not be present in the
database - this is currently not tracked since the functionality is not
commonly used - with path-based vertex id generation, we'll be making
guesses however where empty lookups become "normal" - the same would
happen for incomplete databases as well.

Author: arnetheduck

execution_chain/db/aristo/aristo_desc/desc_structural.nim

@@ -90,8 +90,9 @@ type
       ## unfriendly to caches
 
 const
-  Leaves* = {AccLeaf, StoLeaf}
-  Branches* = {Branch, ExtBranch}
+  Leaves* = {VertexType.AccLeaf, VertexType.StoLeaf}
+  Branches* = {VertexType.Branch, VertexType.ExtBranch}
+  VertexTypes* = Leaves + Branches
 
 # ------------------------------------------------------------------------------
 # Public helpers (misc)

execution_chain/db/aristo/aristo_init/memory_db.nim

@@ -208,7 +208,7 @@ func memoryBackend*(): AristoDbRef =
 
 iterator walkVtx*(
     be: MemBackendRef;
-    kinds = {Branch, ExtBranch, AccLeaf, StoLeaf};
+    kinds = VertexTypes;
       ): tuple[rvid: RootedVertexID, vtx: VertexRef] =
   ##  Iteration over the vertex sub-table.
   for n,rvid in be.sTab.keys.toSeq.sorted:

execution_chain/db/aristo/aristo_init/rocks_db.nim

@@ -245,7 +245,7 @@ proc rocksDbBackend*(
 
 iterator walkVtx*(
     be: RdbBackendRef;
-    kinds = {Branch, ExtBranch, AccLeaf, StoLeaf};
+    kinds = VertexTypes;
       ): tuple[evid: RootedVertexID, vtx: VertexRef] =
   ## Variant of `walk()` iteration over the vertex sub-table.
   for (rvid, vtx) in be.rdb.walkVtx(kinds):

execution_chain/db/aristo/aristo_init/rocks_db/rdb_desc.nim

@@ -73,14 +73,20 @@ type
     Account
     World
 
+  RdbVertexType* = enum
+    Empty
+    Leaf
+    Branch
+    ExtBranch
+
 var
   # Hit/miss counters for LRU cache - global so as to integrate easily with
   # nim-metrics and `uint64` to ensure that increasing them is fast - collection
   # happens from a separate thread.
   # TODO maybe turn this into more general framework for LRU reporting since
   #      we have lots of caches of this sort
   rdbBranchLruStats*: array[RdbStateType, RdbLruCounter]
-  rdbVtxLruStats*: array[RdbStateType, array[VertexType, RdbLruCounter]]
+  rdbVtxLruStats*: array[RdbStateType, array[RdbVertexType, RdbLruCounter]]
   rdbKeyLruStats*: array[RdbStateType, RdbLruCounter]
 
 # ------------------------------------------------------------------------------
@@ -93,6 +99,12 @@ template toOpenArray*(xid: AdminTabID): openArray[byte] =
 template to*(v: RootedVertexID, T: type RdbStateType): RdbStateType =
   if v.root == VertexID(1): RdbStateType.World else: RdbStateType.Account
 
+template to*(v: VertexType, T: type RdbVertexType): RdbVertexType =
+  case v
+  of VertexType.AccLeaf, VertexType.StoLeaf: RdbVertexType.Leaf
+  of VertexType.Branch: RdbVertexType.Branch
+  of VertexType.ExtBranch: RdbVertexType.ExtBranch
+
 template inc*(v: var RdbLruCounter, hit: bool) =
   discard v[hit].fetchAdd(1, moRelaxed)
 

execution_chain/db/aristo/aristo_init/rocks_db/rdb_get.nim

@@ -63,7 +63,7 @@ when defined(metrics):
     # We don't care about synchronization between each type of metric or between
     # the metrics thread and others since small differences like this don't matter
     for state in RdbStateType:
-      for vtype in VertexType:
+      for vtype in RdbVertexType:
         for hit in [false, true]:
           output(
             name = "aristo_rdb_vtx_lru_total",
@@ -198,7 +198,9 @@ proc getVtx*(
         rdb.rdVtxLru.get(rvid.vid)
 
     if rc.isOk:
-      rdbVtxLruStats[rvid.to(RdbStateType)][rc.value().vType].inc(true)
+      rdbVtxLruStats[rvid.to(RdbStateType)][rc.value().vType.to(RdbVertexType)].inc(
+        true
+      )
       return ok(move(rc.value))
 
   # Otherwise fetch from backend database
@@ -214,8 +216,7 @@ proc getVtx*(
     return err((errSym, error))
 
   if not gotData:
-    # As a hack, we count missing data as leaf nodes
-    rdbVtxLruStats[rvid.to(RdbStateType)][VertexType.StoLeaf].inc(false)
+    rdbVtxLruStats[rvid.to(RdbStateType)][RdbVertexType.Empty].inc(false)
     return ok(VertexRef(nil))
 
   if res.isErr():
@@ -224,7 +225,9 @@ proc getVtx*(
   if res.value.vType == Branch:
     rdbBranchLruStats[rvid.to(RdbStateType)].inc(false)
   else:
-    rdbVtxLruStats[rvid.to(RdbStateType)][res.value().vType].inc(false)
+    rdbVtxLruStats[rvid.to(RdbStateType)][res.value().vType.to(RdbVertexType)].inc(
+      false
+    )
 
   # Update cache and return - in peek mode, avoid evicting cache items
   if GetVtxFlag.PeekCache notin flags:

execution_chain/db/aristo/aristo_init/rocks_db/rdb_init.nim

@@ -27,7 +27,7 @@ proc dumpCacheStats(keySize, vtxSize, branchSize: int) =
     echo "vtxLru(", vtxSize, ")"
     echo "   state    vtype       miss        hit      total hitrate"
     for state in RdbStateType:
-      for vtype in VertexType:
+      for vtype in RdbVertexType:
         let
           (miss, hit) = (
             rdbVtxLruStats[state][vtype].get(false),

execution_chain/db/aristo/aristo_walk/memory_only.nim

@@ -29,7 +29,7 @@ export
 iterator walkVtxBe*[T: MemBackendRef](
    _: type T;
    db: AristoDbRef;
-   kinds = {Branch, ExtBranch, AccLeaf, StoLeaf};
+   kinds = VertexTypes;
      ): tuple[rvid: RootedVertexID, vtx: VertexRef] =
   ## Iterate over filtered memory backend or backend-less vertices. This
   ## function depends on the particular backend type name which must match

execution_chain/db/aristo/aristo_walk/persistent.nim

@@ -34,7 +34,7 @@ export
 iterator walkVtxBe*[T: RdbBackendRef](
    _: type T;
    db: AristoDbRef;
-   kinds = {Branch, ExtBranch, AccLeaf, StoLeaf};
+   kinds = VertexTypes;
      ): tuple[rvid: RootedVertexID, vtx: VertexRef] =
   ## Iterate over RocksDB backend vertices. This function depends on
   ## the particular backend type name which must match the backend descriptor.