HA Cluster: INSERT with >5 new property keys per transaction fails with "Error on updating dictionary"

[babasaikiran]

ArcadeDB Version

26.4.2 (build ef892b4/1776994711893/main)

Environment

• 3-node HA cluster (arcadedb.ha.enabled: true, quorum: majority)
• Server mode: test

Expected behavior

INSERT INTO Type CONTENT {"key_0": "v", "key_1": "v", ... "key_49": "v"} should succeed
regardless of how many new (previously unseen) property keys are in the record.

Actual behavior

When a single INSERT introduces more than ~5 new property keys that don't yet exist in the
schema dictionary, the transaction fails with:

Error on transaction commit: Error on updating dictionary for key 'key_N'
com.arcadedb.exception.SchemaException

Steps to reproduce

The script below creates a fresh database, tests INSERT with increasing numbers of new keys,
and reports pass/fail. No external dependencies beyond Python 3 standard library.

Save as dict_bug_repro.py and run:

python dict_bug_repro.py --url http://<arcade-host>:2480/api/v1 --user root --password <password>

#!/usr/bin/env python3
"""
ArcadeDB HA Dictionary Bug Reproducer

Tests INSERT with increasing numbers of new property keys per transaction.
On HA clusters, fails at ~10 new keys per transaction.
"""
import argparse
import json
import time
import urllib.request
import urllib.error
import base64
import sys

DB_NAME = "dict_bug_repro"


def parse_args():
    p = argparse.ArgumentParser(description="ArcadeDB HA dictionary bug reproducer")
    p.add_argument("--url", required=True, help="ArcadeDB HTTP endpoint (e.g. http://host:2480/api/v1)")
    p.add_argument("--user", required=True)
    p.add_argument("--password", required=True)
    return p.parse_args()


class ArcadeClient:
    def __init__(self, url, user, password):
        self.url = url.rstrip("/")
        creds = base64.b64encode(f"{user}:{password}".encode()).decode()
        self.headers = {"Content-Type": "application/json", "Authorization": f"Basic {creds}"}

    def server_command(self, cmd):
        data = json.dumps({"command": cmd}).encode()
        req = urllib.request.Request(f"{self.url}/server", data=data, headers=self.headers, method="POST")
        try:
            with urllib.request.urlopen(req, timeout=30) as r:
                return r.status, r.read().decode()
        except urllib.error.HTTPError as e:
            return e.code, e.read().decode()

    def sql(self, db, command):
        data = json.dumps({"language": "sql", "command": command}).encode()
        req = urllib.request.Request(f"{self.url}/command/{db}", data=data, headers=self.headers, method="POST")
        try:
            with urllib.request.urlopen(req, timeout=30) as r:
                return r.status, r.read().decode()
        except urllib.error.HTTPError as e:
            return e.code, e.read().decode()

    def server_info(self):
        req = urllib.request.Request(f"{self.url}/server", headers=self.headers)
        with urllib.request.urlopen(req, timeout=10) as r:
            return json.loads(r.read().decode())


def main():
    args = parse_args()
    client = ArcadeClient(args.url, args.user, args.password)

    # Server info
    print("Server Info")
    print("-" * 40)
    try:
        info = client.server_info()
        settings = {s["key"]: s["value"] for s in info.get("settings", [])}
        version = info.get("version", "?")
        ha = settings.get("arcadedb.ha.enabled", "?")
        quorum = settings.get("arcadedb.ha.quorum", "?")
        print(f"  Version:    {version}")
        print(f"  HA enabled: {ha}")
        print(f"  HA quorum:  {quorum}")
    except Exception as e:
        print(f"  Failed: {e}")
        sys.exit(1)

    # Setup
    client.server_command(f"drop database {DB_NAME}")
    status, _ = client.server_command(f"create database {DB_NAME}")
    if status != 200:
        print(f"Failed to create database")
        sys.exit(1)
    print(f"\nDatabase '{DB_NAME}' created\n")

    results = []

    # Test 1: INSERT with N new keys
    print("Test 1: INSERT with N new property keys (one transaction)")
    print("-" * 55)
    for n in [1, 2, 3, 5, 8, 10, 15, 20, 50]:
        client.sql(DB_NAME, "DROP TYPE T IF EXISTS UNSAFE")
        client.sql(DB_NAME, "CREATE VERTEX TYPE T IF NOT EXISTS")
        record = {f"key_{i}": f"value_{i}" for i in range(n)}
        status, body = client.sql(DB_NAME, f"INSERT INTO T CONTENT {json.dumps(record)}")
        passed = status == 200
        print(f"  {n:3d} new keys: {'PASS' if passed else 'FAIL'}")
        if not passed:
            detail = json.loads(body).get("detail", "") if body.startswith("{") else body
            print(f"      Error: {str(detail)[:120]}")
        results.append((f"INSERT {n} new keys", passed))
        client.sql(DB_NAME, "DROP TYPE T IF EXISTS UNSAFE")

    # Test 2: 50 keys one at a time (1 key per transaction)
    print(f"\nTest 2: 50 sequential INSERTs (1 new key each)")
    print("-" * 55)
    client.sql(DB_NAME, "CREATE VERTEX TYPE T IF NOT EXISTS")
    ok = 0
    for i in range(50):
        status, _ = client.sql(DB_NAME, f"INSERT INTO T CONTENT {json.dumps({f'prop_{i}': f'v_{i}'})}")
        if status == 200:
            ok += 1
    print(f"  {ok}/50 passed")
    results.append(("50 sequential 1-key INSERTs", ok == 50))
    client.sql(DB_NAME, "DROP TYPE T IF EXISTS UNSAFE")

    # Summary
    print(f"\nSummary")
    print("=" * 55)
    for name, passed in results:
        print(f"  {'PASS' if passed else 'FAIL'}  {name}")

    # Cleanup
    client.server_command(f"drop database {DB_NAME}")
    print(f"\nDatabase '{DB_NAME}' dropped. Done.")


if __name__ == "__main__":
    main()

Sample output (bug confirmed)

Server Info
----------------------------------------
  Version:    26.4.2
  HA enabled: True
  HA quorum:  majority

Database 'dict_bug_repro' created

Test 1: INSERT with N new property keys (one transaction)
-------------------------------------------------------
    1 new keys: PASS
    2 new keys: PASS
    3 new keys: PASS
    5 new keys: PASS
    8 new keys: PASS
   10 new keys: FAIL
      Error: Error on updating dictionary for key 'key_9'
   15 new keys: FAIL
   20 new keys: FAIL
   50 new keys: FAIL

Test 2: 50 sequential INSERTs (1 new key each)
-------------------------------------------------------
  50/50 passed


Summary
=======================================================
  PASS  INSERT 1 new keys
  PASS  INSERT 2 new keys
  PASS  INSERT 3 new keys
  PASS  INSERT 5 new keys
  PASS  INSERT 8 new keys
  FAIL  INSERT 10 new keys
  FAIL  INSERT 15 new keys
  FAIL  INSERT 20 new keys
  FAIL  INSERT 50 new keys
  PASS  50 sequential 1-key INSERTs

[lvca]

Reproduced and fixed.