Skip to content
Cant run my fastapi server on railway
spotcalendar
OP

2 years ago

trying to run the prompt chunker service on railway but it fails to start,

Solved

23 Replies

spotcalendar
OP

2 years ago

ccb57578-3010-4835-bb4c-2018b5b7f683


2 years ago

it's likely that you are running into the 500mb memory limit on the Trial plan


spotcalendar
OP

2 years ago

so i should upgrade it ig


spotcalendar
OP

2 years ago

to 5$ plan


spotcalendar
OP

2 years ago

i hae one query which is

how is more fees added afterwards ?


2 years ago

With the hobby plan, you get $5 worth of usage included every month. You will pay for the amount of resources used during the billing cycle (one month), you only pay $5 (the cost of the hobby plan) if you usage is <= $5, think of it as a $5 minimum spend.


spotcalendar
OP

2 years ago

i boguth the hobby plan


spotcalendar
OP

2 years ago

seeing if this works now


2 years ago

You will need to redeploy all your services for the new limits to apply


spotcalendar
OP

2 years ago

in local takes 68s

1359588107069947904


spotcalendar
OP

2 years ago

68ms


spotcalendar
OP

2 years ago

but on railway same thing takes 6-7s


spotcalendar
OP

2 years ago

any reasons why ?


spotcalendar
OP

2 years ago

i m have put like time logs, will be able to share that in a moment


2 years ago

looks like you're running something that uses GPU?


spotcalendar
OP

2 years ago

yes


spotcalendar
OP

2 years ago

sentence transformers


spotcalendar
OP

2 years ago

but small text like these shouldn't take 7s


spotcalendar
OP

2 years ago

ig


2 years ago

there are no GPUs on Railway, you would be running the embeddings on CPU when on Railway


spotcalendar
OP

2 years ago

so there is no way to solve the latency for this ?


2 years ago

run a more CPU efficient model, or use an external AI service


2 years ago

!s


Status changed to Solved brody • over 1 year ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...