ludo-web-multiplayer/troubleshooting.md

9.6 KiB
Raw Permalink Blame History

Youve identified the Mongo volume thats causing the auth issue.

What this means local mern-ludo_mongo_data

This volume still exists, so Mongo is:

reusing old, broken initialization

not recreating the admin user

continuing to fail authentication

We now need to delete this volume so Mongo can initialize cleanly.

Do this EXACTLY (safe + correct) 🔴 Step 1: Stop containers docker compose down

🔴 Step 2: Remove the Mongo volume docker volume rm mern-ludo_mongo_data

If Docker complains its in use, run:

docker compose down -v

(that removes volumes automatically)

Step 3: Confirm its gone docker volume ls | grep ludo

👉 This should return nothing.

Step 4: Start ONLY Mongo first (important) docker compose up -d mongo

Wait 1015 seconds.

Check logs:

docker logs ludo-mongo --tail=30

You should NOT see:

bad file

Unable to acquire security key

crash/restart loops

Step 5: Test authentication (this MUST work) docker exec -it ludo-mongo mongosh -u admin -p adminpassword --authenticationDatabase admin

If you get a > prompt → 🎉 auth is fixed

Step 6: Initialize replica set (ONCE)

Inside mongosh:

rs.initiate({ _id: "rs0", members: [{ _id: 0, host: "mongo:27017" }] })

Then:

rs.status().members[0].stateStr

Must be:

"PRIMARY"

Step 7: Start the app docker compose up -d app

(or just)

docker compose up -d

Then:

docker restart mern-ludo-app

🎯 Expected final result Mongo

Auth works

Replica set PRIMARY

No errors in logs

Backend MongoDB connected Server listening on port 18081

Browser

App loads

No crashes

No noscript / auth errors

🧾 One-line status

The Mongo volume still existed; deleting it allows Mongo to create the admin user correctly.

Once Step 5 succeeds, you are fully unblocked. Run Step 2 now and tell me when mongosh login works 👍

Youre very close, and the remaining issue is not random.

What the two messages really mean (important) 1 These messages are NOT the main problem Collection [local.oplog.rs] not found ReadConcernMajorityNotAvailableYet

These appear when:

replica set is not initiated yet

PRIMARY is not elected yet

They are expected noise before rs.initiate() succeeds.

👉 Well fix them automatically once the replica set is initialized.

2 THIS is the real blocker 🔴 MongoServerError: Authentication failed

This means 100%:

🔐 MongoDB is running with authentication enabled, but the admin user does not exist.

This is why:

backend cant connect

mongosh -u admin fails

oplog cant be created

replica set cant complete

Why the admin user does NOT exist (root cause)

Mongo creates the root user only once, on first successful startup, when:

/data/db is empty

Mongo reaches a stable running state

Earlier, Mongo:

crashed due to keyFile issues

restarted several times

skipped user creation

but auth stayed enabled

Result:

Mongo requires auth, but no users exist → permanent auth failure

This is a classic Mongo + replica set + keyFile edge case.

The ONLY correct fix (clean & guaranteed)

We must let Mongo start once without auth, create the user manually, then re-enable auth.

This is the official MongoDB recovery method.

🔥 DO THIS STEP-BY-STEP (no skipping) STEP 1 Stop everything docker compose down

STEP 2 Temporarily DISABLE auth (Mongo only)

Edit docker-compose.yml mongo service:

🔴 TEMPORARY CHANGE

Comment out these lines for now:

environment:

MONGO_INITDB_ROOT_USERNAME: admin

MONGO_INITDB_ROOT_PASSWORD: adminpassword

And change the command to NO keyFile, NO auth:

command:

  • mongod
  • --replSet
  • rs0
  • --bind_ip_all

(Leave volumes as-is.)

STEP 3 Start Mongo only docker compose up -d mongo

Wait ~10 seconds.

STEP 4 Connect WITHOUT auth (this must work) docker exec -it ludo-mongo mongosh

If you get a prompt → good.

STEP 5 Create admin user MANUALLY

Inside mongosh:

use admin db.createUser({ user: "admin", pwd: "adminpassword", roles: [ { role: "root", db: "admin" } ] })

You must see:

{ ok: 1 }

STEP 6 Initialize replica set (NOW it will work) rs.initiate({ _id: "rs0", members: [{ _id: 0, host: "mongo:27017" }] })

Wait ~510 seconds, then:

rs.status().members[0].stateStr

Must be:

"PRIMARY"

Exit mongosh:

exit

STEP 7 Re-enable auth + keyFile (FINAL config)

Restore your correct mongo config:

environment: MONGO_INITDB_ROOT_USERNAME: admin MONGO_INITDB_ROOT_PASSWORD: adminpassword

command:

  • bash
  • -c
  • | cp /tmp/mongo-keyfile /data/db/keyfile && chown mongodb:mongodb /data/db/keyfile && chmod 400 /data/db/keyfile && exec mongod --replSet rs0 --keyFile /data/db/keyfile --bind_ip_all

STEP 8 Restart clean docker compose down docker compose up -d

STEP 9 Test auth (this WILL work now) docker exec -it ludo-mongo mongosh -u admin -p adminpassword --authenticationDatabase admin

STEP 🔟 Start / restart backend docker restart mern-ludo-app

What you should see now Mongo

Auth works

PRIMARY elected

local.oplog.rs exists

No auth errors

Backend MongoDB connected Server listening on port 18081

App

Loads normally

No crashes

Change Streams work

🧠 One-line final diagnosis

MongoDB authentication was enabled before any user existed; manual user creation was required to recover.

This is the last MongoDB trap — once done, you will not hit this again.

MongoServerError[NotWritablePrimary]: not primary

Perfect — that error is expected at this exact moment, and it tells us precisely what to do next 👍 You didnt break anything.

What this error means (very clear) MongoServerError[NotWritablePrimary]: not primary

This means:

MongoDB is running as a replica set member, but no PRIMARY has been elected yet.

Until a node becomes PRIMARY:

you cannot create users

you cannot write data

db.createUser() will fail

So this is not an auth issue anymore.

Correct order (important)

When running Mongo as a replica set, the order must be:

Start Mongo

Initiate replica set

Wait until PRIMARY

Create users

Enable auth (already planned)

Right now, youre at step 2½.

DO THIS NOW (inside mongosh)

You are already connected with:

docker exec -it ludo-mongo mongosh

At the > prompt, run THIS FIRST 👇

🔥 Step 1: Initiate replica set rs.initiate({ _id: "rs0", members: [ { _id: 0, host: "mongo:27017" } ] })

Expected output { ok: 1 }

🔥 Step 2: Wait a few seconds, then verify PRIMARY rs.status().members[0].stateStr

You must see:

"PRIMARY"

If it still says STARTUP or SECONDARY, wait 5 seconds and run again.

ONLY AFTER PRIMARY → create admin user

Now run:

use admin

db.createUser({ user: "admin", pwd: "adminpassword", roles: [ { role: "root", db: "admin" } ] })

Expected: { ok: 1 }

🧠 Why this happened (important lesson)

You started Mongo with:

mongod --replSet rs0

That automatically puts Mongo into replica set mode, but:

replica set is inactive

no PRIMARY exists yet

Mongo rejects all writes until a PRIMARY is elected.

This is correct MongoDB behavior.

After user creation (next steps preview)

Once { ok: 1 } appears:

Exit mongosh:

exit

Re-enable auth + keyFile in docker-compose (as discussed)

Restart containers

Verify:

docker exec -it ludo-mongo mongosh -u admin -p adminpassword --authenticationDatabase admin

🎯 One-line status update

MongoDB is in replica-set mode but not PRIMARY yet — initialize the replica set first.

👉 Action for you now

Run rs.initiate(...), wait for PRIMARY, then retry db.createUser.

Tell me what rs.status().members[0].stateStr prints, and well finish this cleanly 💪

version: "3.9"

services:
  mongo:
    image: mongo:7.0
    container_name: ludo-mongo
    restart: unless-stopped
    networks:
      - ludo-net
    environment:
      MONGO_INITDB_ROOT_USERNAME: admin
      MONGO_INITDB_ROOT_PASSWORD: adminpassword
    volumes:
      - ludo-mongo-data:/data/db
      - ./mongo-keyfile:/tmp/mongo-keyfile:ro
    command:
      - bash
      - -c
      - |
        cp /tmp/mongo-keyfile /data/db/keyfile &&
        chown mongodb:mongodb /data/db/keyfile &&
        chmod 400 /data/db/keyfile &&
        exec mongod --replSet rs0 --keyFile /data/db/keyfile --bind_ip_all        
    ports:
      - "27017:27017"

  app:
    build: .
    container_name: mern-ludo-app
    restart: unless-stopped
    networks:
      - ludo-net
    env_file:
      - .env
    ports:
      - "18081:8080"
    depends_on:
      - mongo

networks:
  ludo-net:

volumes:
  ludo-mongo-data:

Step 2 Recreate keyfile cleanly (one last time) cd ~/mern-ludo rm -f mongo-keyfile openssl rand -base64 756 > mongo-keyfile chmod 400 mongo-keyfile

Verify:

wc -c mongo-keyfile

Step 3 Full reset (MANDATORY) docker compose down -v docker compose up -d --build

Wait ~10 seconds.

Step 4 Confirm Mongo is RUNNING (not restarting) docker ps

You should NOT see Restarting.

Step 5 Now exec WILL work docker exec -it ludo-mongo mongosh -u admin -p adminpassword --authenticationDatabase admin

If you get a shell → 🎉 Mongo is fixed

Step 6 Initialize replica set (ONCE) rs.initiate({ _id: "rs0", members: [{ _id: 0, host: "mongo:27017" }] })

Check:

rs.status().members[0].stateStr // PRIMARY

Step 7 Restart app docker restart mern-ludo-app

Your Change Stream error will now be gone permanently.

🧠 Why ALL previous attempts failed Attempt Why it failed Bind mount keyfile Ownership mismatch chmod 400 only Not enough DietPi ARM Stricter FS behavior Mongo 7 Enforces keyfile rules hard

This copy + chown pattern solves all of them.

One-line final diagnosis

MongoDB rejected the keyfile because bind-mounted files keep host ownership, and Mongo runs as a non-root user.