ryyst/kalzu-value-store
1
0
Fork 1
Code Issues 5 Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request 'Added Merkel Trees for better replication state tracking.' (#1) from MrKalzu/kalzu-value-store:master into master

Browse Source 
Reviewed-on: #1
This commit is contained in:
ryyst
2025-09-11 17:55:59 +03:00
parent ebed73dc11 9ea19a3532
commit 3775939a3b
2 changed files with 1035 additions and 429 deletions
Download Patch File Download Diff File Expand all files Collapse all files

117
integration_test.sh
View File

@ -1,7 +1,7 @@
#!/bin/bash #!/bin/bash
# KVS Integration Test Suite - Working Version # KVS Integration Test Suite - Adapted for Merkle Tree Sync
# Tests all critical features of the distributed key-value store # Tests all critical features of the distributed key-value store with Merkle Tree replication
# Colors for output # Colors for output
RED='\033[0;31m' RED='\033[0;31m'
@ -43,7 +43,7 @@ test_start() {
# Cleanup function # Cleanup function
cleanup() { cleanup() {
log_info "Cleaning up test environment..." log_info "Cleaning up test environment..."
pkill -f "./kvs" 2>/dev/null || true pkill -f "$BINARY" 2>/dev/null || true
rm -rf "$TEST_DIR" 2>/dev/null || true rm -rf "$TEST_DIR" 2>/dev/null || true
sleep 2 # Allow processes to fully terminate sleep 2 # Allow processes to fully terminate
} }
@ -75,6 +75,7 @@ test_build() {
log_error "Binary build failed" log_error "Binary build failed"
return 1 return 1
fi fi
# Ensure we are back in TEST_DIR for subsequent tests
cd "$TEST_DIR" cd "$TEST_DIR"
} }
@ -103,12 +104,12 @@ EOF
-d '{"message":"hello world"}') -d '{"message":"hello world"}')
local get_result=$(curl -s http://localhost:8090/kv/test/basic) local get_result=$(curl -s http://localhost:8090/kv/test/basic)
local message=$(echo "$get_result" | jq -r '.message' 2>/dev/null) local message=$(echo "$get_result" | jq -r '.data.message' 2>/dev/null) # Adjusted jq path
if [ "$message" = "hello world" ]; then if [ "$message" = "hello world" ]; then
log_success "Basic CRUD operations work" log_success "Basic CRUD operations work"
else else
log_error "Basic CRUD failed: $get_result" log_error "Basic CRUD failed: Expected 'hello world', got '$message' from $get_result"
fi fi
else else
log_error "Basic test node failed to start" log_error "Basic test node failed to start"
@ -120,7 +121,7 @@ EOF
# Test 3: Cluster formation # Test 3: Cluster formation
test_cluster_formation() { test_cluster_formation() {
test_start "2-node cluster formation" test_start "2-node cluster formation and Merkle Tree replication"
# Node 1 config # Node 1 config
cat > cluster1.yaml <<EOF cat > cluster1.yaml <<EOF
@ -158,7 +159,7 @@ EOF
return 1 return 1
fi fi
sleep 2 sleep 2 # Give node 1 a moment to fully initialize
$BINARY cluster2.yaml >/dev/null 2>&1 & $BINARY cluster2.yaml >/dev/null 2>&1 &
local pid2=$! local pid2=$!
@ -168,28 +169,46 @@ EOF
return 1 return 1
fi fi
# Wait for cluster formation # Wait for cluster formation and initial Merkle sync
sleep 8 sleep 15
# Check if nodes see each other # Check if nodes see each other
local node1_members=$(curl -s http://localhost:8101/members/ | jq length 2>/dev/null || echo 0) local node1_members=$(curl -s http://localhost:8101/members/ | jq length 2>/dev/null || echo 0)
local node2_members=$(curl -s http://localhost:8102/members/ | jq length 2>/dev/null || echo 0) local node2_members=$(curl -s http://localhost:8102/members/ | jq length 2>/dev/null || echo 0)
if [ "$node1_members" -ge 1 ] && [ "$node2_members" -ge 1 ]; then if [ "$node1_members" -ge 1 ] && [ "$node2_members" -ge 1 ]; then
log_success "2-node cluster formed successfully" log_success "2-node cluster formed successfully (N1 members: $node1_members, N2 members: $node2_members)"
# Test data replication # Test data replication
log_info "Putting data on Node 1, waiting for Merkle sync..."
curl -s -X PUT http://localhost:8101/kv/cluster/test \ curl -s -X PUT http://localhost:8101/kv/cluster/test \
-H "Content-Type: application/json" \ -H "Content-Type: application/json" \
-d '{"source":"node1"}' >/dev/null -d '{"source":"node1", "value": 1}' >/dev/null
sleep 12 # Wait for sync cycle # Wait for Merkle sync cycle to complete
sleep 12
local node2_data=$(curl -s http://localhost:8102/kv/cluster/test | jq -r '.source' 2>/dev/null) local node2_data_full=$(curl -s http://localhost:8102/kv/cluster/test)
if [ "$node2_data" = "node1" ]; then local node2_data_source=$(echo "$node2_data_full" | jq -r '.data.source' 2>/dev/null)
log_success "Data replication works correctly" local node2_data_value=$(echo "$node2_data_full" | jq -r '.data.value' 2>/dev/null)
local node1_data_full=$(curl -s http://localhost:8101/kv/cluster/test)
if [ "$node2_data_source" = "node1" ] && [ "$node2_data_value" = "1" ]; then
log_success "Data replication works correctly (Node 2 has data from Node 1)"
# Verify UUIDs and Timestamps are identical (crucial for Merkle sync correctness)
local node1_uuid=$(echo "$node1_data_full" | jq -r '.uuid' 2>/dev/null)
local node1_timestamp=$(echo "$node1_data_full" | jq -r '.timestamp' 2>/dev/null)
local node2_uuid=$(echo "$node2_data_full" | jq -r '.uuid' 2>/dev/null)
local node2_timestamp=$(echo "$node2_data_full" | jq -r '.timestamp' 2>/dev/null)
if [ "$node1_uuid" = "$node2_uuid" ] && [ "$node1_timestamp" = "$node2_timestamp" ]; then
log_success "Replicated data retains original UUID and Timestamp"
else
log_error "Replicated data changed UUID/Timestamp: N1_UUID=$node1_uuid, N1_TS=$node1_timestamp, N2_UUID=$node2_uuid, N2_TS=$node2_timestamp"
fi
else else
log_error "Data replication failed: $node2_data" log_error "Data replication failed: Node 2 data: $node2_data_full"
fi fi
else else
log_error "Cluster formation failed (N1 members: $node1_members, N2 members: $node2_members)" log_error "Cluster formation failed (N1 members: $node1_members, N2 members: $node2_members)"
@ -199,9 +218,12 @@ EOF
sleep 2 sleep 2
} }
# Test 4: Conflict resolution (simplified) # Test 4: Conflict resolution (Merkle Tree based)
# This test assumes 'test_conflict.go' creates two BadgerDBs with a key
# that has the same path and timestamp but different UUIDs, or different timestamps
# but same path. The Merkle tree sync should then trigger conflict resolution.
test_conflict_resolution() { test_conflict_resolution() {
test_start "Conflict resolution test" test_start "Conflict resolution test (Merkle Tree based)"
# Create conflicting data using our utility # Create conflicting data using our utility
rm -rf conflict1_data conflict2_data 2>/dev/null || true rm -rf conflict1_data conflict2_data 2>/dev/null || true
@ -233,7 +255,8 @@ sync_interval: 8
EOF EOF
# Start nodes # Start nodes
$BINARY conflict1.yaml >conflict1.log 2>&1 & # Node 1 started first, making it "older" for tie-breaker if timestamps are equal
"$BINARY" conflict1.yaml >conflict1.log 2>&1 &
local pid1=$! local pid1=$!
if wait_for_service 8111; then if wait_for_service 8111; then
@ -242,26 +265,50 @@ EOF
local pid2=$! local pid2=$!
if wait_for_service 8112; then if wait_for_service 8112; then
# Get initial data # Get initial data (full StoredValue)
local node1_initial=$(curl -s http://localhost:8111/kv/test/conflict/data | jq -r '.message' 2>/dev/null) local node1_initial_full=$(curl -s http://localhost:8111/kv/test/conflict/data)
local node2_initial=$(curl -s http://localhost:8112/kv/test/conflict/data | jq -r '.message' 2>/dev/null) local node2_initial_full=$(curl -s http://localhost:8112/kv/test/conflict/data)
# Wait for conflict resolution local node1_initial_msg=$(echo "$node1_initial_full" | jq -r '.data.message' 2>/dev/null)
sleep 12 local node2_initial_msg=$(echo "$node2_initial_full" | jq -r '.data.message' 2>/dev/null)
# Get final data log_info "Initial conflict state: Node1='$node1_initial_msg', Node2='$node2_initial_msg'"
local node1_final=$(curl -s http://localhost:8111/kv/test/conflict/data | jq -r '.message' 2>/dev/null)
local node2_final=$(curl -s http://localhost:8112/kv/test/conflict/data | jq -r '.message' 2>/dev/null) # Wait for conflict resolution (multiple sync cycles might be needed)
sleep 20
# Get final data (full StoredValue)
local node1_final_full=$(curl -s http://localhost:8111/kv/test/conflict/data)
local node2_final_full=$(curl -s http://localhost:8112/kv/test/conflict/data)
local node1_final_msg=$(echo "$node1_final_full" | jq -r '.data.message' 2>/dev/null)
local node2_final_msg=$(echo "$node2_final_full" | jq -r '.data.message' 2>/dev/null)
# Check if they converged # Check if they converged
if [ "$node1_final" = "$node2_final" ] && [ -n "$node1_final" ]; then if [ "$node1_final_msg" = "$node2_final_msg" ] && [ -n "$node1_final_msg" ]; then
if grep -q "conflict resolution" conflict1.log conflict2.log 2>/dev/null; then log_success "Conflict resolution converged to: '$node1_final_msg'"
log_success "Conflict resolution detected and resolved ($node1_initial vs $node2_initial → $node1_final)"
# Verify UUIDs and Timestamps are identical after resolution
local node1_final_uuid=$(echo "$node1_final_full" | jq -r '.uuid' 2>/dev/null)
local node1_final_timestamp=$(echo "$node1_final_full" | jq -r '.timestamp' 2>/dev/null)
local node2_final_uuid=$(echo "$node2_final_full" | jq -r '.uuid' 2>/dev/null)
local node2_final_timestamp=$(echo "$node2_final_full" | jq -r '.timestamp' 2>/dev/null)
if [ "$node1_final_uuid" = "$node2_final_uuid" ] && [ "$node1_final_timestamp" = "$node2_final_timestamp" ]; then
log_success "Resolved data retains consistent UUID and Timestamp across nodes"
else else
log_success "Nodes converged without conflicts ($node1_final)" log_error "Resolved data has inconsistent UUID/Timestamp: N1_UUID=$node1_final_uuid, N1_TS=$node1_final_timestamp, N2_UUID=$node2_final_uuid, N2_TS=$node2_final_timestamp"
fi fi
# Optionally, check logs for conflict resolution messages
if grep -q "Conflict resolved" conflict1.log conflict2.log 2>/dev/null; then
log_success "Conflict resolution messages found in logs"
else
log_error "No 'Conflict resolved' messages found in logs, but data converged."
fi
else else
log_error "Conflict resolution failed: N1='$node1_final', N2='$node2_final'" log_error "Conflict resolution failed: N1_final='$node1_final_msg', N2_final='$node2_final_msg'"
fi fi
else else
log_error "Conflict node 2 failed to start" log_error "Conflict node 2 failed to start"
@ -276,14 +323,14 @@ EOF
sleep 2 sleep 2
else else
cd "$TEST_DIR" cd "$TEST_DIR"
log_error "Failed to create conflict test data" log_error "Failed to create conflict test data. Ensure test_conflict.go is correct."
fi fi
} }
# Main test execution # Main test execution
main() { main() {
echo "==================================================" echo "=================================================="
echo " KVS Integration Test Suite" echo " KVS Integration Test Suite (Merkle Tree)"
echo "==================================================" echo "=================================================="
# Setup # Setup
@ -308,7 +355,7 @@ main() {
echo "==================================================" echo "=================================================="
if [ $TESTS_FAILED -eq 0 ]; then if [ $TESTS_FAILED -eq 0 ]; then
echo -e "${GREEN}🎉 All tests passed! KVS is working correctly.${NC}" echo -e "${GREEN}🎉 All tests passed! KVS with Merkle Tree sync is working correctly.${NC}"
cleanup cleanup
exit 0 exit 0
else else

991
main.go
View File

File diff suppressed because it is too large Load Diff