Merge pull request #4839 from NixOS/ca/gracefully-handle-duplicate-realisations

Gracefully handle duplicate realisations

src/libstore/build/drv-output-substitution-goal.cc

@@ -17,6 +17,13 @@ DrvOutputSubstitutionGoal::DrvOutputSubstitutionGoal(const DrvOutput& id, Worker
 void DrvOutputSubstitutionGoal::init()
 {
     trace("init");
+
+    /* If the derivation already exists, we’re done */
+    if (worker.store.queryRealisation(id)) {
+        amDone(ecSuccess);
+        return;
+    }
+
     subs = settings.useSubstitutes ? getDefaultSubstituters() : std::list<ref<Store>>();
     tryNext();
 }
@@ -53,9 +60,23 @@ void DrvOutputSubstitutionGoal::tryNext()
         return;
     }
 
-    for (const auto & [drvOutputDep, _] : outputInfo->dependentRealisations) {
-        if (drvOutputDep != id) {
-            addWaitee(worker.makeDrvOutputSubstitutionGoal(drvOutputDep));
+    for (const auto & [depId, depPath] : outputInfo->dependentRealisations) {
+        if (depId != id) {
+            if (auto localOutputInfo = worker.store.queryRealisation(depId);
+                localOutputInfo && localOutputInfo->outPath != depPath) {
+                warn(
+                    "substituter '%s' has an incompatible realisation for '%s', ignoring.\n"
+                    "Local:  %s\n"
+                    "Remote: %s",
+                    sub->getUri(),
+                    depId.to_string(),
+                    worker.store.printStorePath(localOutputInfo->outPath),
+                    worker.store.printStorePath(depPath)
+                );
+                tryNext();
+                return;
+            }
+            addWaitee(worker.makeDrvOutputSubstitutionGoal(depId));
         }
     }
 

src/libstore/local-store.cc

@@ -53,6 +53,7 @@ struct LocalStore::State::Stmts {
     SQLiteStmt InvalidatePath;
     SQLiteStmt AddDerivationOutput;
     SQLiteStmt RegisterRealisedOutput;
+    SQLiteStmt UpdateRealisedOutput;
     SQLiteStmt QueryValidDerivers;
     SQLiteStmt QueryDerivationOutputs;
     SQLiteStmt QueryRealisedOutput;
@@ -345,6 +346,15 @@ LocalStore::LocalStore(const Params & params)
                 values (?, ?, (select id from ValidPaths where path = ?), ?)
                 ;
             )");
+        state->stmts->UpdateRealisedOutput.create(state->db,
+            R"(
+                update Realisations
+                    set signatures = ?
+                where
+                    drvPath = ? and
+                    outputName = ?
+                ;
+            )");
         state->stmts->QueryRealisedOutput.create(state->db,
             R"(
                 select Realisations.id, Output.path, Realisations.signatures from Realisations
@@ -710,14 +720,46 @@ void LocalStore::registerDrvOutput(const Realisation & info)
     settings.requireExperimentalFeature("ca-derivations");
     retrySQLite<void>([&]() {
         auto state(_state.lock());
+        if (auto oldR = queryRealisation_(*state, info.id)) {
+            if (info.isCompatibleWith(*oldR)) {
+                auto combinedSignatures = oldR->signatures;
+                combinedSignatures.insert(info.signatures.begin(),
+                    info.signatures.end());
+                state->stmts->UpdateRealisedOutput.use()
+                    (concatStringsSep(" ", combinedSignatures))
+                    (info.id.strHash())
+                    (info.id.outputName)
+                    .exec();
+            } else {
+                throw Error("Trying to register a realisation of '%s', but we already "
+                            "have another one locally.\n"
+                            "Local:  %s\n"
+                            "Remote: %s",
+                    info.id.to_string(),
+                    printStorePath(oldR->outPath),
+                    printStorePath(info.outPath)
+                );
+            }
+        } else {
             state->stmts->RegisterRealisedOutput.use()
                 (info.id.strHash())
                 (info.id.outputName)
                 (printStorePath(info.outPath))
                 (concatStringsSep(" ", info.signatures))
                 .exec();
+        }
         uint64_t myId = state->db.getLastInsertedRowId();
-        for (auto & [outputId, _] : info.dependentRealisations) {
+        for (auto & [outputId, depPath] : info.dependentRealisations) {
+            auto localRealisation = queryRealisationCore_(*state, outputId);
+            if (!localRealisation)
+                throw Error("unable to register the derivation '%s' as it "
+                            "depends on the non existent '%s'",
+                    info.id.to_string(), outputId.to_string());
+            if (localRealisation->second.outPath != depPath)
+                throw Error("unable to register the derivation '%s' as it "
+                            "depends on a realisation of '%s' that doesn’t"
+                            "match what we have locally",
+                    info.id.to_string(), outputId.to_string());
             state->stmts->AddRealisationReference.use()
                 (myId)
                 (outputId.strHash())
@@ -1734,12 +1776,12 @@ void LocalStore::createUser(const std::string & userName, uid_t userId)
     }
 }
 
-std::optional<const Realisation> LocalStore::queryRealisation(
-    const DrvOutput& id) {
-    typedef std::optional<const Realisation> Ret;
-    return retrySQLite<Ret>([&]() -> Ret {
-        auto state(_state.lock());
-        auto useQueryRealisedOutput(state->stmts->QueryRealisedOutput.use()
+std::optional<std::pair<int64_t, Realisation>> LocalStore::queryRealisationCore_(
+        LocalStore::State & state,
+        const DrvOutput & id)
+{
+    auto useQueryRealisedOutput(
+            state.stmts->QueryRealisedOutput.use()
                 (id.strHash())
                 (id.outputName));
     if (!useQueryRealisedOutput.next())
@@ -1749,31 +1791,53 @@ std::optional<const Realisation> LocalStore::queryRealisation(
     auto signatures =
         tokenizeString<StringSet>(useQueryRealisedOutput.getStr(2));
 
-        std::map<DrvOutput, StorePath> dependentRealisations;
-        auto useRealisationRefs(
-            state->stmts->QueryRealisationReferences.use()
-                (realisationDbId));
-        while (useRealisationRefs.next()) {
-            auto depHash = useRealisationRefs.getStr(0);
-            auto depOutputName = useRealisationRefs.getStr(1);
-            auto useQueryRealisedOutput(state->stmts->QueryRealisedOutput.use()
-                    (depHash)
-                    (depOutputName));
-            assert(useQueryRealisedOutput.next());
-            auto outputPath = parseStorePath(useQueryRealisedOutput.getStr(1));
-            auto depId = DrvOutput { Hash::parseAnyPrefixed(depHash), depOutputName };
-            dependentRealisations.insert({depId, outputPath});
-        }
-
-        return Ret{Realisation{
+    return {{
+        realisationDbId,
+        Realisation{
             .id = id,
             .outPath = outputPath,
             .signatures = signatures,
-            .dependentRealisations = dependentRealisations,
+        }
     }};
-    });
 }
 
+std::optional<const Realisation> LocalStore::queryRealisation_(
+            LocalStore::State & state,
+            const DrvOutput & id)
+{
+    auto maybeCore = queryRealisationCore_(state, id);
+    if (!maybeCore)
+        return std::nullopt;
+    auto [realisationDbId, res] = *maybeCore;
+
+    std::map<DrvOutput, StorePath> dependentRealisations;
+    auto useRealisationRefs(
+        state.stmts->QueryRealisationReferences.use()
+            (realisationDbId));
+    while (useRealisationRefs.next()) {
+        auto depId = DrvOutput {
+            Hash::parseAnyPrefixed(useRealisationRefs.getStr(0)),
+            useRealisationRefs.getStr(1),
+        };
+        auto dependentRealisation = queryRealisationCore_(state, depId);
+        assert(dependentRealisation); // Enforced by the db schema
+        auto outputPath = dependentRealisation->second.outPath;
+        dependentRealisations.insert({depId, outputPath});
+    }
+
+    res.dependentRealisations = dependentRealisations;
+
+    return { res };
+}
+
+std::optional<const Realisation>
+LocalStore::queryRealisation(const DrvOutput & id)
+{
+    return retrySQLite<std::optional<const Realisation>>([&]() {
+        auto state(_state.lock());
+        return queryRealisation_(*state, id);
+    });
+}
 
 FixedOutputHash LocalStore::hashCAPath(
     const FileIngestionMethod & method, const HashType & hashType,

src/libstore/local-store.hh

@@ -203,6 +203,8 @@ public:
     void registerDrvOutput(const Realisation & info, CheckSigsFlag checkSigs) override;
     void cacheDrvOutputMapping(State & state, const uint64_t deriver, const string & outputName, const StorePath & output);
 
+    std::optional<const Realisation> queryRealisation_(State & state, const DrvOutput & id);
+    std::optional<std::pair<int64_t, Realisation>> queryRealisationCore_(State & state, const DrvOutput & id);
     std::optional<const Realisation> queryRealisation(const DrvOutput&) override;
 
 private:

src/libstore/realisation.cc

@@ -140,6 +140,16 @@ StorePath RealisedPath::path() const {
     return std::visit([](auto && arg) { return arg.getPath(); }, raw);
 }
 
+bool Realisation::isCompatibleWith(const Realisation & other) const
+{
+    assert (id == other.id);
+    if (outPath == other.outPath) {
+        assert(dependentRealisations == other.dependentRealisations);
+        return true;
+    }
+    return false;
+}
+
 void RealisedPath::closure(
     Store& store,
     const RealisedPath::Set& startPaths,

src/libstore/realisation.hh

@@ -47,6 +47,8 @@ struct Realisation {
     static std::set<Realisation> closure(Store &, const std::set<Realisation> &);
     static void closure(Store &, const std::set<Realisation> &, std::set<Realisation>& res);
 
+    bool isCompatibleWith(const Realisation & other) const;
+
     StorePath getPath() const { return outPath; }
 
     GENERATE_CMP(Realisation, me->id, me->outPath);

tests/ca/duplicate-realisation-in-closure.sh

@@ -0,0 +1,26 @@
+source ./common.sh
+
+sed -i 's/experimental-features .*/& ca-derivations ca-references/' "$NIX_CONF_DIR"/nix.conf
+
+export REMOTE_STORE_DIR="$TEST_ROOT/remote_store"
+export REMOTE_STORE="file://$REMOTE_STORE_DIR"
+
+rm -rf $REMOTE_STORE_DIR
+clearStore
+
+# Build dep1 and push that to the binary cache.
+# This entails building (and pushing) current-time.
+nix copy --to "$REMOTE_STORE" -f nondeterministic.nix dep1
+clearStore
+sleep 2 # To make sure that `$(date)` will be different
+# Build dep2.
+# As we’ve cleared the cache, we’ll have to rebuild current-time. And because
+# the current time isn’t the same as before, this will yield a new (different)
+# realisation
+nix build -f nondeterministic.nix dep2
+
+# Build something that depends both on dep1 and dep2.
+# If everything goes right, we should rebuild dep2 rather than fetch it from
+# the cache (because that would mean duplicating `current-time` in the closure),
+# and have `dep1 == dep2`.
+nix build --substituters "$REMOTE_STORE" -f nondeterministic.nix toplevel --no-require-sigs

tests/ca/nondeterministic.nix

@@ -0,0 +1,35 @@
+with import ./config.nix;
+
+let mkCADerivation = args: mkDerivation ({
+    __contentAddressed = true;
+    outputHashMode = "recursive";
+    outputHashAlgo = "sha256";
+} // args);
+in
+
+rec {
+  currentTime = mkCADerivation {
+    name = "current-time";
+    buildCommand = ''
+      mkdir $out
+      echo $(date) > $out/current-time
+    '';
+  };
+  dep = seed: mkCADerivation {
+    name = "dep";
+    inherit seed;
+    buildCommand = ''
+      echo ${currentTime} > $out
+    '';
+  };
+  dep1 = dep 1;
+  dep2 = dep 2;
+  toplevel = mkCADerivation {
+    name = "toplevel";
+    buildCommand = ''
+      test ${dep1} == ${dep2}
+      touch $out
+    '';
+  };
+}
+

tests/local.mk

@@ -47,6 +47,7 @@ nix_tests = \
   compute-levels.sh \
   ca/build.sh \
   ca/build-with-garbage-path.sh \
+  ca/duplicate-realisation-in-closure.sh \
   ca/substitute.sh \
   ca/signatures.sh \
   ca/nix-shell.sh \