Troubleshooting

This guide helps you diagnose and resolve common issues when working with Memori. For additional support, visit our Discord community or GitHub Issues.

Installation Issues

ModuleNotFoundError: No module named ‘memori’

Python cannot find the Memori module

Problem: Import fails with ModuleNotFoundError.Symptoms:

>>> from memori import Memori
ModuleNotFoundError: No module named 'memori'

Solutions:

Verify installation

pip list | grep memori

If not listed, install it:

pip install memori

Check Python environment

Ensure you’re using the correct Python interpreter:

which python
python --version

If using virtual environments:

source venv/bin/activate  # Linux/Mac
venv\Scripts\activate     # Windows
pip install memori

Reinstall if necessary

pip uninstall memori
pip install memori

Missing Dependencies for Database Drivers

ImportError: No module named 'psycopg2' or similar

Problem: Database driver not installed.Symptoms:

ImportError: No module named 'psycopg2'
ImportError: No module named 'pymongo'
ImportError: No module named 'pymysql'

Solutions:Memori has minimal dependencies by default. Install database drivers as needed:

# PostgreSQL (psycopg2)
pip install psycopg2-binary

# PostgreSQL (psycopg3) - recommended for CockroachDB
pip install "memori[cockroachdb]"
# or manually:
pip install "psycopg[binary]>=3.1.0"

# MongoDB
pip install pymongo

# MySQL/MariaDB
pip install pymysql

# Oracle
pip install oracledb

All drivers at once (development):

pip install psycopg2-binary psycopg[binary] pymongo pymysql

SSL Certificate Errors

SSL: CERTIFICATE_VERIFY_FAILED

Problem: SSL verification fails when connecting to Memori API or databases.Symptoms:

SSLError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed

Solutions:

Update certificates (recommended):

# macOS
pip install --upgrade certifi

# Ubuntu/Debian
sudo apt-get update
sudo apt-get install ca-certificates

# RHEL/CentOS
sudo yum update ca-certificates

Corporate proxy/firewall: Configure SSL verification:

import os
os.environ['REQUESTS_CA_BUNDLE'] = '/path/to/ca-bundle.crt'

Development only - disable verification (NOT for production):
```
import urllib3
urllib3.disable_warnings()
```

Configuration Issues

API Key Not Recognized

MEMORI_API_KEY environment variable issues

Problem: API key not being recognized or applied.Symptoms:

Still hitting IP-based quota limits
Authentication errors
Quota command shows lower limits than expected

Solutions:

Verify API key is set

echo $MEMORI_API_KEY

Should output your API key. If empty, set it:

export MEMORI_API_KEY="your_api_key_here"

Check for whitespace

Remove any extra whitespace:

export MEMORI_API_KEY=$(echo $MEMORI_API_KEY | tr -d '[:space:]')

Restart your application

Environment variables are read at startup. Restart your application after setting the key.

Verify in code

import os
print(f"API Key: {os.getenv('MEMORI_API_KEY')}")

Make persistent (Linux/Mac)

Add to your shell profile:

echo 'export MEMORI_API_KEY="your_key"' >> ~/.bashrc
source ~/.bashrc

Connection String Issues

Database connection string errors

Problem: Cannot connect to database using connection string.Symptoms:

OperationalError: could not connect to server
ConnectionError: connection refused

Solutions:

Verify connection string format:

# PostgreSQL
"postgresql://user:password@host:5432/database"

# CockroachDB
"postgresql://user:password@host:26257/defaultdb?sslmode=verify-full"

# MongoDB
"mongodb://user:password@host:27017/database"

# MySQL
"mysql://user:password@host:3306/database"

Check URL encoding for special characters:

from urllib.parse import quote_plus

password = "p@ssw0rd!#$"
encoded_password = quote_plus(password)
connection_string = f"postgresql://user:{encoded_password}@host/db"

Test connectivity separately:

# PostgreSQL
psql "postgresql://user:password@host:5432/database"

# MongoDB
mongosh "mongodb://user:password@host:27017/database"

Verify network access:

# Test if port is accessible
nc -zv host 5432  # PostgreSQL
nc -zv host 26257 # CockroachDB
nc -zv host 27017 # MongoDB

Embedding Model Download Failures

Cannot download sentence-transformers models

Problem: Embedding model download fails or times out.Symptoms:

OSError: Can't load model
HTTPError: 503 Server Error

Solutions:

Manual download with retry:
```
python -m memori setup
```

Configure cache directory:

export TRANSFORMERS_CACHE="/path/to/cache"
export SENTENCE_TRANSFORMERS_HOME="/path/to/cache"
python -m memori setup

Use offline mode (pre-download):

from sentence_transformers import SentenceTransformer

# Download on machine with internet
model = SentenceTransformer("all-MiniLM-L6-v2")
model.save("/path/to/model")

# Load on offline machine
model = SentenceTransformer("/path/to/model")

Check firewall/proxy settings:

export HTTP_PROXY="http://proxy:8080"
export HTTPS_PROXY="http://proxy:8080"
python -m memori setup

Runtime Issues

Memory Not Being Recalled

LLM interactions not recalling previous context

Problem: Memori not recalling memories from previous interactions.Symptoms:

LLM doesn’t remember previous conversations
No context from past sessions
Memories created but not retrieved

Solutions:

Verify attribution is set

from memori import Memori
from openai import OpenAI

mem = Memori()
client = mem.llm.register(OpenAI())

# REQUIRED: Set attribution before interactions
mem.attribution(entity_id="user_123", process_id="agent")

# Verify it's set
print(f"Entity: {mem.config.entity_id}")
print(f"Process: {mem.config.process_id}")
print(f"Session: {mem.config.session_id}")

Check recall configuration

# Ensure recall limits are reasonable
print(f"Embeddings limit: {mem.config.recall_embeddings_limit}")
print(f"Facts limit: {mem.config.recall_facts_limit}")
print(f"Relevance threshold: {mem.config.recall_relevance_threshold}")

# Increase if too restrictive
mem.config.recall_embeddings_limit = 2000
mem.config.recall_relevance_threshold = 0.05

Verify memories are being created

Check the dashboard at app.memorilabs.ai to see if memories are being stored.

Check quota status

python -m memori quota

If quota is exceeded, new memories won’t be created.

Slow Performance

High latency or slow response times

Problem: Memori operations are slower than expected.Symptoms:

High latency (>500ms) for memory recall
Slow LLM response times
Application feels sluggish

Solutions:

Optimize recall settings:

mem.config.recall_embeddings_limit = 500  # Reduce search space
mem.config.recall_facts_limit = 3         # Fewer facts
mem.config.recall_relevance_threshold = 0.2  # Stricter filtering

Use faster embedding model:

export MEMORI_EMBEDDINGS_MODEL="all-MiniLM-L6-v2"
python -m memori setup

Check database performance (BYODB):

-- PostgreSQL: Check slow queries
SELECT query, mean_exec_time
FROM pg_stat_statements
ORDER BY mean_exec_time DESC
LIMIT 10;

-- Ensure indexes exist
\di memories

Use connection pooling:

import psycopg2.pool

pool = psycopg2.pool.ThreadedConnectionPool(
    minconn=5,
    maxconn=20,
    host="localhost",
    database="memori"
)

mem = Memori().storage.register(pool.getconn())

Profile your code:

import time

start = time.time()
response = client.chat.completions.create(...)
print(f"LLM call: {time.time() - start:.3f}s")

See Performance Tuning for more optimization strategies.

Memory Leaks

Application memory usage growing over time

Problem: Python process memory usage continuously increases.Symptoms:

High RAM usage
OOM (Out of Memory) errors after running for extended periods
Slow degradation of performance

Solutions:

Reduce thread pool size:

from concurrent.futures import ThreadPoolExecutor
mem.config.thread_pool_executor = ThreadPoolExecutor(max_workers=5)

Explicitly close connections:

# For BYODB
connection = psycopg2.connect(...)
mem = Memori().storage.register(connection)

# When done
connection.close()

Process in batches:

for batch in chunked(items, 100):
    process_batch(batch)
    # Reset cache periodically
    mem.config.reset_cache()

Monitor with memory profiler:

pip install memory-profiler
python -m memory_profiler your_script.py

LLM Provider Issues

OpenAI Integration Issues

OpenAI client not working with Memori

Problem: OpenAI client fails or doesn’t integrate properly.Symptoms:

AttributeError: 'OpenAI' object has no attribute '_memori'
TypeError: unsupported operand type(s)

Solutions:

Verify OpenAI version:

pip show openai
# Requires openai>=1.0.0
pip install --upgrade openai

Correct initialization order:

from memori import Memori
from openai import OpenAI

# Correct: Register client with Memori
client = OpenAI()
mem = Memori().llm.register(client)

# Set attribution before use
mem.attribution(entity_id="user", process_id="agent")

# Now use client
response = client.chat.completions.create(...)

Check API key:
```
echo $OPENAI_API_KEY
```

Test without Memori first:

from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "test"}]
)
print(response.choices[0].message.content)

Anthropic Integration Issues

Anthropic Claude client issues

Problem: Anthropic client not working with Memori.Solutions:

Verify Anthropic SDK version:

pip show anthropic
pip install --upgrade anthropic

Correct usage:

from memori import Memori
from anthropic import Anthropic

client = Anthropic()
mem = Memori().llm.register(client)
mem.attribution(entity_id="user_123", process_id="claude_agent")

response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}]
)

Check API key:
```
echo $ANTHROPIC_API_KEY
```

Streaming Not Working

Streaming responses fail or are incomplete

Problem: Streaming LLM responses don’t work properly with Memori.Symptoms:

Stream stops prematurely
Incomplete responses
Memory not recorded for streamed content

Solutions:

Ensure proper iteration:

stream = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Tell a story"}],
    stream=True
)

# Iterate through ALL chunks
for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

# Memory recorded after stream completes

Don’t manually close stream early:

# Wrong: Closing stream early
for i, chunk in enumerate(stream):
    if i > 10:
        break  # Memory won't be recorded!

# Correct: Consume entire stream
for chunk in stream:
    process(chunk)

Async streaming:

from openai import AsyncOpenAI

client = AsyncOpenAI()
mem = Memori().llm.register(client)

stream = await client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello"}],
    stream=True
)

async for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Database-Specific Issues

PostgreSQL Connection Issues

Cannot connect to PostgreSQL/CockroachDB

Problem: PostgreSQL or CockroachDB connection fails.Common errors:

psycopg2.OperationalError: could not connect to server
psycopg2.OperationalError: FATAL: password authentication failed
ssl.SSLError: [SSL: CERTIFICATE_VERIFY_FAILED]

Solutions:

Verify connection parameters:

import psycopg2

connection = psycopg2.connect(
    host="localhost",
    port=5432,
    database="memori",
    user="postgres",
    password="your_password",
    sslmode="prefer"  # or "require" for CockroachDB
)

Test connection separately:

psql -h localhost -p 5432 -U postgres -d memori

CockroachDB SSL requirements:

# CockroachDB requires SSL
connection = psycopg2.connect(
    "postgresql://user:password@host:26257/defaultdb?sslmode=verify-full"
)

Check firewall:

# Test port accessibility
nc -zv localhost 5432

Verify PostgreSQL is running:

# Linux
sudo systemctl status postgresql

# macOS
brew services list | grep postgresql

# Docker
docker ps | grep postgres

MongoDB Connection Issues

Cannot connect to MongoDB

Problem: MongoDB connection fails.Common errors:

pymongo.errors.ServerSelectionTimeoutError
pymongo.errors.ConfigurationError

Solutions:

Verify connection string:

from pymongo import MongoClient

# Standard MongoDB
client = MongoClient("mongodb://localhost:27017/")

# MongoDB Atlas
client = MongoClient(
    "mongodb+srv://user:[email protected]/?retryWrites=true&w=majority"
)

# With authentication
client = MongoClient(
    host="localhost",
    port=27017,
    username="user",
    password="password",
    authSource="admin"
)

Test connection:
```
mongosh "mongodb://localhost:27017"
```

Check MongoDB is running:

# Linux
sudo systemctl status mongod

# macOS
brew services list | grep mongodb

# Docker
docker ps | grep mongo

Verify network access:
```
nc -zv localhost 27017
```

SQLite Permission Issues

SQLite database permission errors

Problem: Cannot create or write to SQLite database.Symptoms:

sqlite3.OperationalError: unable to open database file
PermissionError: [Errno 13] Permission denied

Solutions:

Check file permissions:

ls -l memori.db
chmod 644 memori.db  # Read/write for owner

Ensure directory exists and is writable:
```
mkdir -p ~/.memori
chmod 755 ~/.memori
```

Use absolute path:

import sqlite3
import os
from memori import Memori

db_path = os.path.expanduser("~/.memori/memori.db")
os.makedirs(os.path.dirname(db_path), exist_ok=True)

connection = sqlite3.connect(db_path)
mem = Memori().storage.register(connection)

Check disk space:
```
df -h .
```

CockroachDB Cluster Management

Cluster Creation Fails

CockroachDB cluster start command fails

Problem: python -m memori cockroachdb cluster start fails.Symptoms:

RuntimeError: the cluster failed to come online
ModuleNotFoundError: No module named 'psycopg'

Solutions:

Install required dependencies:

pip install "memori[cockroachdb]"
# or manually:
pip install "psycopg[binary]>=3.1.0"

Wait longer (can take 2+ minutes): The cluster creation process can take several minutes. Be patient.
Check network connectivity:
```
curl -I https://api.memorilabs.ai
```
Verify API access:
```
python -m memori quota
```
Try again: If it fails, the cluster is automatically cleaned up. Simply retry:
```
python -m memori cockroachdb cluster start
```

Cannot Access Cluster

Lost connection string or cannot access cluster

Problem: Lost CockroachDB connection string or cannot connect.Solutions:

Important: Memori Labs cannot recover lost connection strings. The connection string is displayed only once during cluster creation for security reasons.

If you lost your connection string:

Delete the old cluster:

python -m memori cockroachdb cluster delete

Create a new cluster:

python -m memori cockroachdb cluster start

Save the connection string immediately in a secure password manager.

Getting Help

Collect Diagnostic Information

When reporting issues, include:

# System information
python --version
pip show memori
pip show openai anthropic google-genai  # If using

# Environment
echo $MEMORI_API_KEY | cut -c1-10  # First 10 chars only
echo $MEMORI_EMBEDDINGS_MODEL

# Test basic functionality
python -m memori quota

# Enable debug logging
export MEMORI_DEBUG=1
python your_script.py 2>&1 | tee debug.log

Enable Debug Logging

import logging

# Enable Memori debug logs
logging.basicConfig(level=logging.DEBUG)
logger = logging.getLogger("memori")
logger.setLevel(logging.DEBUG)

# Your code here

Common Error Messages

Error Message	Likely Cause	Quick Fix
`ModuleNotFoundError: No module named 'memori'`	Not installed	`pip install memori`
`No attribution set`	Missing attribution	`mem.attribution(entity_id=..., process_id=...)`
`could not connect to server`	Database unreachable	Check connection string, network
`CERTIFICATE_VERIFY_FAILED`	SSL certificate issue	Update certificates
`quota exceeded`	Hit quota limit	Sign up for API key or upgrade
`No module named 'psycopg'`	Missing driver	`pip install memori[cockroachdb]`

Support Channels

Discord Community

Get help from the community and Memori team

GitHub Issues

Report bugs and request features

Documentation

Browse complete documentation

Email Support

Contact the Memori team directly

Next Steps

CLI Usage

Master the Memori CLI

Performance Tuning

Optimize your deployment

Quota Management

Monitor and manage quota

Contributing

Contribute to Memori

Use Cases

Examples

Advanced

Contributing

​Installation Issues

​ModuleNotFoundError: No module named ‘memori’

​Missing Dependencies for Database Drivers

​SSL Certificate Errors

​Configuration Issues

​API Key Not Recognized

​Connection String Issues

​Embedding Model Download Failures

​Runtime Issues

​Memory Not Being Recalled

​Slow Performance

​Memory Leaks

​LLM Provider Issues

​OpenAI Integration Issues

​Anthropic Integration Issues

​Streaming Not Working

​Database-Specific Issues

​PostgreSQL Connection Issues

​MongoDB Connection Issues

​SQLite Permission Issues

​CockroachDB Cluster Management

​Cluster Creation Fails

​Cannot Access Cluster

​Getting Help

​Collect Diagnostic Information

​Enable Debug Logging

​Common Error Messages

​Support Channels

Discord Community

GitHub Issues

Documentation

Email Support

​Next Steps

CLI Usage

Performance Tuning

Quota Management

Contributing

Build docs developers (and LLMs) love

Installation Issues

ModuleNotFoundError: No module named ‘memori’

Missing Dependencies for Database Drivers

SSL Certificate Errors

Configuration Issues

API Key Not Recognized

Connection String Issues

Embedding Model Download Failures

Runtime Issues

Memory Not Being Recalled

Slow Performance

Memory Leaks

LLM Provider Issues

OpenAI Integration Issues

Anthropic Integration Issues

Streaming Not Working

Database-Specific Issues

PostgreSQL Connection Issues

MongoDB Connection Issues

SQLite Permission Issues

CockroachDB Cluster Management

Cluster Creation Fails

Cannot Access Cluster

Getting Help

Collect Diagnostic Information

Enable Debug Logging

Common Error Messages

Support Channels

Next Steps