Go to your Redis service settings > Networking. If there are any capitals in the URL, try changing it to all lowercase then redeploy your Redis and backend service.

Additionally, make sure that both services are in the same project and environment.

..

tried that, there is no capitals in the URL.

this is the error on my backend log:

[Nest] 13 - 02/12/2026, 4:00:54 PM ERROR [ConsultationLiveUpdatesService] Reader error: getaddrinfo ENOTFOUND redis.railway.internal

[Nest] 13 - 02/12/2026, 4:00:54 PM ERROR [ConsultationLiveUpdatesService] Subscriber error: getaddrinfo ENOTFOUND redis.railway.internal

[Nest] 13 - 02/12/2026, 4:00:56 PM WARN [CacheService] Redis connection failed: getaddrinfo ENOTFOUND redis.railway.internal. Cache operations will be disabled.

[Nest] 13 - 02/12/2026, 4:00:56 PM ERROR [ConsultationLiveUpdatesService] Reader error: getaddrinfo ENOTFOUND redis.railway.internal

Are your backend and Redis in the same project and running in the same environment?

I think there is a general issue with internal url - I am also not able to connect to postgrees instances.

The status page is not still mentioning anything - I think the team should investigate what is happening.

 Your backend logs show it's connecting to redis.railway.internal but your actual Redis hostname is redis-xjxw.railway.internal.Looks like the URL in your backend config is missing the -xjxw suffix.

Mine doesn't have it, should he remove the suffix instead?

The hostname depends on your service name in railway. The safest approach is to not hardcode it at all. Go to your Redis service tab copy the exact REDIS_URL value, or use ${{Redis.REDIS_URL}} as a variable reference in your backend service. That way it always uses whatever Railway assigns, suffix or not.

You can also just edit the private networking URL and/or variable directly.

Later changed it to redis.railway.internal, but still does not work

Tried this as well, but it still does not work

Can you try connecting to redis using the external/public url (TCP proxy) as a temporary workaround? Go to your rredis service -> Settings ->

networking -> enable TCP proxy, then use that url instead.

Note that this will incur egress fees.

let me try

Nope it still does not work

Sorry man, I couldn’t be of that much help. If I were you, I’d delete and recreate the redis service for a clean start. Just keep in mind this will wipe any data stored in redis. If you’re only using it for caching and pub/sub, that’s fine; otherwise, make sure to export the data first.

But I’m curious like what’s the main issue here? could it be something internal on railway?

Can you share a screenshot of your service canvas?

Tried that as well, it doesnt work

Here it is

The env var name which my backend uses is REDIS_URL or REDIS_HOST
but i in the Redis service the name of vars there is like REDISURL or REDISHOST
and i am trying to reference it in my backend

this could be a problem?

Maybe

It seems like you generated a domain.

You need to generate a TCP proxy. Not a domain.

but this will incur egress fees?

This is for testing purposes, but yes you are correct. Using public networking will incur egress fees.

also TCP proxy is only needed if you want to connect to Redis from outside Railway right?

Yes.

Services such as Postgres or Redis require TCP proxies for public networking, not domains.

 Can you check your variable names?

these are some variables i am using
on backend which is perfectly has the values of redis
REDIS_DB

*******

REDIS_HOST

*******

REDIS_KEY_PREFIX

*******

REDIS_PASSWORD

*******

REDIS_PORT

*******

REDIS_TTL

*******

REDIS_URL

I tried deleting the domain, and deployed a tcp proxy and use that as my REDIS URL
it does not work gives these errors:

[Nest] 13 - 02/13/2026, 3:49:51 AM ERROR [ConsultationLiveUpdatesService] Subscriber error: getaddrinfo ENOTFOUND redis.railway.internal

[Nest] 13 - 02/13/2026, 3:49:51 AM ERROR [ConsultationLiveUpdatesService] Reader error: getaddrinfo ENOTFOUND redis.railway.internal

[Nest] 13 - 02/13/2026, 3:49:51 AM WARN [CacheService] Redis connection failed: getaddrinfo ENOTFOUND redis.railway.internal. Cache operations will be disabled.

SIGTERM received, shutting down gracefully

[Nest] 13 - 02/13/2026, 3:49:51 AM LOG [ConsultationLiveUpdatesService] Closing Redis subscriber connection...

[Nest] 13 - 02/13/2026, 3:49:51 AM LOG [ConsultationLiveUpdatesService] Closing Redis reader connection...

But are the values set manually or using railway variable references? can you check does REDIS_HOST contain  something like redis.railway.internal (hardcoded) or ${{Redis.REDISHOST}} (reference)?

You're still using the *.railway.internal URL.

I'd just try SSH-ing into the container and pinging your TCP proxy for now to see if it can resolve.

How do i ssh?

Right click your service and click Copy SSH Command.

You'll need Railway CLI installed prior to this.

i tried with the proxy url, still it uses *.railway.internal

Just curious is hostname hardcoded somewhere in your code?

i tried this says: proxy is open

no it is not

proxy is open means the SSH tunnel is connected. now in that same terminal, can you try:

redis-cli -h localhost -p <the-port-shown> ping

yeah it pongs back

so redis is good, issue is 100% your backend not connecting to it correctly.. Can you search your entire codebase for railway.internal and also check if you have a  .env file that might be overriding Railway's environment variables? (I know you already said you didn't hardcode but can you check once more )

checked again, it is not hardcoded anywhere i added a log in the backend which runs after successful deployment:
[Nest] 13 - 02/13/2026, 4:48:47 AM LOG [CacheService] Redis config mode=url (source=REDIS_URL, nodeEnv=production, url=redis://default:***@redis-stage.railway.internal:6379)

but still get the same subscriber and reader error:
[Nest] 13 - 02/13/2026, 4:49:43 AM ERROR [ConsultationLiveUpdatesService] Reader error: getaddrinfo ENOTFOUND redis-stage.railway.internal

[Nest] 13 - 02/13/2026, 4:49:43 AM ERROR [ConsultationLiveUpdatesService] Subscriber error: getaddrinfo ENOTFOUND redis-stage.railway.internal

[Nest] 13 - 02/13/2026, 4:49:45 AM WARN [CacheService] Redis connection failed: getaddrinfo ENOTFOUND redis-stage.railway.internal. Cache operations will be disabled

i've recently changed the name to redis-stage to check

the weired part is this error is only on staging environment
the connection for backend and redis works perfectly on prod

I feel this is almost certainly an env issue. The REDIS_URL is resolving to redis-stage.railway.internal but private networking DNS can't find it.Is your Redis service actually deployed in the staging environment? Go to the Redis service -> make sure you're looking at the staging

environment tab and it shows as active.
=
If Redis was only deployed to production but not staging, the redis-stage.railway.internal hostname simply doesn't exist in the staging network hence ENOTFOUND. Also  did the redis service get redeployed after you renamed it to redis-stage?

Also  did the redis service get redeployed after you renamed it to redis-stage?
=> yes

I feel this is almost certainly an env issue. The REDIS_URL is resolving to redis-stage.railway.internal but private networking DNS can't find it.Is your Redis service actually deployed in the staging environment? Go to the Redis service -> make sure you're looking at the staging

=> yes, right environment, deployment is active, vars are referenced, same project, same env, same region

I have've exhausted all the usual suspects :( wondering if railway's internal DNS for the staging environment has a cached/broken record that didn't update after the rename or redeploy .

Makes sense, how can we talk to the railway support team for this issue?

how do i raise the railway support ticket? for this ?
this is a major outage

Exact problem:

• Staging backend couldn’t resolve Redis internal hostname (ENOTFOUND redis-*.railway.internal).

• Our backend uses ioredis, which in some Railway legacy/private-network cases can fail with IPv4-only DNS lookup.

• Because Redis was unreachable, session cache checks failed, causing 401 Session expired or invalid.

What I fixed:

• Updated backend Redis clients to force dual-stack DNS lookup (family: 0) in both places:

  • cache client (CacheService)

  • live updates clients (ConsultationLiveUpdatesService)

• This makes hostname resolution work across IPv4/IPv6 environments.

Result:

• Redis connectivity path is now production-safe for both staging and prod-style environments

so it's IPv6 vs IPv4!