minor flume
Trying to add a healthcheck to a service but for some reason I'm getting the following issue, seems like it's not actually a health issue with my service but with Railway itself? The error seems off, never got that one specifically.
Healthcheck works just fine when trying it through the browser
A environment variable for the port is set
minor flume
project id: 6097cd13-bfe2-485b-ba57-61aebc55857a
service id: c25cda1d-e54b-4de9-8d8d-8e37db3b7753
deployment id: a4ac502a-4462-4b18-b4f2-5607d4ab484b
grim peak
is hyperdx listening on ipv4?
minor flume
uhmm i've no idea
let me see if I can check the code, there's no env variable for it afaik
oh it uses nextjs under the hood so probably that env variable you told me should work
I tried to set the HOSTNAME=0.0.0.0 and still no luck
grim peak
?
minor flume
ok I'm going crazy, sometimes it works, sometimes it does not?
like freshly new created project, same env variables
minor flume
like it seems to work when I poke around with it, maybe if I remove the healthcheck to allow for an active deploy and then add it again it works?
i'll confirm
steps to reproduce:
1 - deploy the following template https://railway.com/deploy/clickstack
2 - go to hyperdx service and see the weird error
3 - remove healthcheck from the service, deploy it
4 - add a healthcheck right after an active deployment
grim peak
thank you, i will pass that info along
grim peak
i've traced our healthcheck code for the v2 runtime, and the Error occurred while trying to proxy: healthcheck.railway.app/health text comes directly from the http request via code that reads directly from the response body of the http health check, this is not text from our code.
minor flume
i've traced our healthcheck code for the v2 runtime, and the `Error occurred whi...
so not an issue with Railway? seems weird that it works on second deployment
grim peak
yep not an issue with our health checks
minor flume
i do not see any string match with Error occurred on hyperdx 🤔
grim peak
likely an upstream error
minor flume
grim peak
and there we go
minor flume
maybe the healthcheck tries to check if it's publicly available and then fails
yeah seems like removing the healthcheck is the only option
grim peak
we do set the host header to healthcheck.railway.app but I'm fairly certain you know that
is there a route that does not use that middleware?
minor flume
/api/clickhouse-proxy uses http-proxy-middleware
ah, one that does not use
maybe I can change the server_url to be localhost, seems like it's not used anywhere else
or even the private domain
grim peak
now you're thinking
minor flume
now its going through, btw maybe that error from the healthcheck should be improved to make it clear that it's from the origin
anyway, i guess thats all, one small problem left is that the OTEL Collector depends on HyperDX but their healthcheck runs on a different port
is it possible to create a deploy/build dependency (that message saying "waiting for deploy..." or smth like that)?
im using referenced variables from HyperDX but it seems that no dependency is created
grim peak
are the services referencing each other? if so, they are deployed at the same time
minor flume
maybe im confusing things up but doesn't railway have an option to wait for a successful deploy when it detects a referenced variable?
grim peak
yes it does, but no waiting will happen if the two services reference each other
minor flume
hmm fair enough
i'll hardcode a specific environment variable, not the best option but until we get healthcheck to specific ports it's fine
nvm seems like i cant do that, they're referencing private domains which may change depending on the service name
i'll leave it as it is
anyway, thread can be closed, thanks for the help brody
grim peak
no problem!
grim peak
!s
vernal basin