Hello @Brody ! Do you reckon there might be something that I might've misconfigured? If it's the case for a bug is there a workaround for it?

please do not tag team or mods - <#727685388893945877> #5

anyway, what makes you say the container isn't being ran?

Thank you for your response. The fact that it is not displaying anything in the Deploy logs section. yet, it should since it thoretcally "has been run"

+1 – I am seeing the same. I waited up to an hour in the event there was some queue delay.

I see several posts that seem to suggest crons have stopped working or there is a behavior change: https://help.railway.app/feedback/cron-reliability-improvements-c2f2c959

im not seeing any cron services in any of your environments?

Um, I think there should be. development-testnet environment Cron Jobs group. I think service id of one of the cron jobs is: 17c81b48-6409-4b66-ba85-5537f35e62aa

I must be blind then, my bad, I'll give another look

Please, it would be really helpful. Maybe there is something misconfigured on our end

I'll look when I'm back at my computer <:salute:1137099685417451530>

but in the mean time, please tell me more about this cron job.

language?

what does it achieve?

are you using a logger library?

etc

update, found the service, something really strange is going on here for sure

Great news. The service is trying to run a Dockerfile that runs a binary file created after bun env build action. Other than that it is pretty standard. Locally (on my machine Macbook M1 series) everything works fine. For some reason on the Railway env nothing works

Language is Typescript

Run time environment is bun.js

logger is used everywhere in our project so I doubt it's the case, we use pino

The railway.toml is provided above. Here is the Dockerfile:

FROM node:20.12.2 AS build

RUN npm install -g bun@1.1.26
WORKDIR /app
COPY . .
RUN bun install --frozen-lockfile --prefer-offline

ARG TURBO_TEAM
ENV TURBO_TEAM=$TURBO_TEAM
ARG TURBO_TOKEN
ENV TURBO_TOKEN=$TURBO_TOKEN
RUN bun turbo build --filter secret-cron-name

FROM gcr.io/distroless/base-debian11
WORKDIR /app
COPY --from=build /app/cron/secret-cron-name/bin /app/bin

CMD [ "/app/bin" ]```

what are your logging settings? are you logging unbuffered to stdout / stderr?

Yes. It worked for a couple of days, now they are not being run again. Seems like the cron jobs are very unreliable

are you sure your application is exiting after it's done it's task?

Yes, it even has a process.exit(0) at the end

when testing locally, it always ends

how long does it take?

1 should take under 30 seconds. The other should probably take 5 minutes initially. Aftewards it will take under 30 seconds as well

what is your cron schedule

0 * * * *

For example, this is the cron service that does not run: b734ede7-3e09-4b94-8699-28d74a046e30

okay so every hour right?

yes

tried running it manually and it still does not log anything

what logging library are you using?

pino - https://www.npmjs.com/package/pino

are you the same person who reported that railway was overwriting the log level?

No, this is the only thread I have opened

okay, thank you.

do you have a custom transport?

No, standard stdout/stderr

I'm fairly confident Pino is buffering logs, if Pino is buffering logs railway can't pick the logs up to send to the frontend

We are using pino for all of the services. They all log perfectly fine. The only service instances that don't print out the logs are the crons. Moreover, some of them should surely not fail, but I can't see their effects. This leads me to thinking that the cron jobs are not run in the first place

does railway's UI show you that your job was ran?

or does it say skipped?

can you enable the beta cron UI in the service settings

Yes, I will attach a video in a bit

Actually, it would take too much to record a video. I will summarize it. Pressing on run now, redeploys the cron job

Nothing else shows up. I will send you a screenshot in 3 minutes when the previous deployment shows that's supposed to run

can you try duplicating the service? (right click -> duplicate)

It should have ran but it didn't

Done, duplicating it

in our backend we have this marked as completed

There's definitely something wrong. I am using this kind of dockerfile

so probably you only mark this as completed because its being built?