Timeouts starting the LibreChat container in slower hosts

[madrover]

Hi
I run Librechat in my Synology DS920+ NAS using Docker Compose and, while it generally works very well, the LibreChat container often struggles to start in a timely manner. It's quite apparent this issue is caused with the slower specs of the NAS plus the slower magnetic HDDs and I solve this by manually restarting the Docker Compose api service a few times until it works.

 > [email protected] backend:dev
 > cross-env NODE_ENV=development npx nodemon api/server/index.js

 npm warn exec The following package was not found and will be installed: [email protected]
 [nodemon] 3.1.9
 [nodemon] to restart at any time, enter `rs`
 [nodemon] watching path(s): *.*
 [nodemon] watching extensions: js,mjs,cjs,json
 [nodemon] starting `node api/server/index.js`
 2025-03-02T22:31:01.181Z info: [Optional] Redis not initialized. Note: Redis support is experimental.
 2025-03-02T22:31:35.921Z error: An error occurred while fixing Token indexes: Operation `tokens.indexes()` buffering timed out after 10000ms
 2025-03-02T22:31:35.929Z error: KeyvMongo connection error: connection timed out
 2025-03-02T22:31:35.931Z error: There was an uncaught error: connection timed out
 2025-03-02T22:31:35.932Z debug: [indexSync] Clearing sync timeouts before exiting...
[nodemon] app crashed - waiting for file changes before starting...

Would it be possible to configure these timeouts to cater lower specced devices? When all services are running everything runs smoothly, it's just the time out in the startup which is problematic.

On the other hand, even though the service in the LibreChat container is effectively crashed, Docker does not detect it so automatic container restarts cannot be triggered. Ideally, this container should have a healtcheck so it can be restarted automatically whenever this misbehavior happens.

Thanks and keep up the good work!

[rubentalstra]

So what are you saying? You're requesting a health check of the container?

[madrover]

What I’m looking for is a configurable timeout setting that can enable more reliable starts on slower systems were slower execution pace is expected. Would that be feasible? Is there any other suitable approach?

Having a healtcheck in the default Compose file or in the Librechat container would be a nice addition, but does not solve per se the startup timeout issue.

Thanks!

[rubentalstra]

@madrover

so something like this you mean?

[rubentalstra]

be060cb

[madrover]

Can’t test right now, as I’m AFK. That’s the idea for a healthcheck, although I’m not 100% sure that CMD is correct. Also, I’m incIined to say that including the healtcheck in the Dockerfile would be nicer, because it would be available if deploying the container to other runtime environments such as Kubernetes, Fargate, etc…

[rubentalstra]

great idea. I have updated it in the Dockerfile cc8e4db

[madrover]

I suggest the following configurations:

    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:${PORT}/health"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 30s

and

HEALTHCHECK --interval=10s --timeout=5s --retries=3 --start-period=30s \
  CMD curl -f http://localhost:${PORT}/health

I think these timeout values are more aligned with the expected application behavior. Also, the || exit 1 was redundant, the -f caters for HTTP errors (4xx/5xx responses) and other connection failures already return non-zero codes that will make the healthcheck fail.

While this is a welcome addition it does not solve the timeout issue. Do you have any suggestion on that front?

Thanks!

[Machiel692]

Hi I run Librechat in my Synology DS920+ NAS using Docker Compose and, while it generally works very well, the LibreChat container often struggles to start in a timely manner. It's quite apparent this issue is caused with the slower specs of the NAS plus the slower magnetic HDDs and I solve this by manually restarting the Docker Compose api service a few times until it works.

 > [email protected] backend:dev
 > cross-env NODE_ENV=development npx nodemon api/server/index.js

 npm warn exec The following package was not found and will be installed: [email protected]
 [nodemon] 3.1.9
 [nodemon] to restart at any time, enter `rs`
 [nodemon] watching path(s): *.*
 [nodemon] watching extensions: js,mjs,cjs,json
 [nodemon] starting `node api/server/index.js`
 2025-03-02T22:31:01.181Z info: [Optional] Redis not initialized. Note: Redis support is experimental.
 2025-03-02T22:31:35.921Z error: An error occurred while fixing Token indexes: Operation `tokens.indexes()` buffering timed out after 10000ms
 2025-03-02T22:31:35.929Z error: KeyvMongo connection error: connection timed out
 2025-03-02T22:31:35.931Z error: There was an uncaught error: connection timed out
 2025-03-02T22:31:35.932Z debug: [indexSync] Clearing sync timeouts before exiting...
[nodemon] app crashed - waiting for file changes before starting...

Would it be possible to configure these timeouts to cater lower specced devices? When all services are running everything runs smoothly, it's just the time out in the startup which is problematic.

On the other hand, even though the service in the LibreChat container is effectively crashed, Docker does not detect it so automatic container restarts cannot be triggered. Ideally, this container should have a healtcheck so it can be restarted automatically whenever this misbehavior happens.

Thanks and keep up the good work!

Based on your experience sofar with LibreChat on Synology NAS, could you share your detailed step-by-step instructions to install LibreChat on the NAS? I have tried multiple ways but sofar I keep getting errors. I'm sure the whole community would appreciate your experience! So many thanks in advance already now!!

[madrover]

The main issue was the mongodb version; you need to use an old version because newer ones require some specific processor capabilities not present in my Synology NAS at least.

Checking out this project and using the following docker-compose.override.yml should be enough to make it work via Docker Compose:

services:
  api:
    command: npm run backend:dev
    volumes:
      - ./librechat.yaml:/app/librechat.yaml
  mongodb:
    image: mongo:4.4.18

With these changes in place just use the regular docker-compose up -d from the shell and you're done.

In general it works very well. Sometimes there's the issue that I've described in this ticket and you need to restart the API a few times, but that's all. Personally, I installed a NVMe drive into my NAS and moved Docker to it, so I'm not hitting it any more. Maybe newer LibreChat versions behave better on this regards.

[Machiel692]

Thanks a lot!! This is very helpful!