a year ago
We had a deployment running succesfully for the last week
today, the railway instances cant communicate with the servers in Hetzner datacenter, and the otherway around is true too
is there some way we can diagnose this?
feels like railway has blocked the whole ip range
1 Replies
https://discord.com/channels/713503345364697088/727689277219012669/1229812857898799104
same message, with extra info
a year ago
Railway does not block anything of this sort unless there was an attack originating from Hetzner
what errors are you getting?
deployments on railway cant access items in hetzner (services)
and hetzner services cant access a DB in railway
a year ago
let keep this chat in here please, chit chat would not be the place
a year ago
please note I don't work for Railway.
can you connect to the database locally?
a year ago
are you sure you are using all the correct credentials?
a year ago
can you ssh into the VPS and do some debugging, pinging, etc
a year ago
let's not jump to such conclusions just yet
a year ago
are you able to ping that host from your vps (the same VPS that is having connection issues)
a year ago
can you use the mysql cli from within the VPS to connect to the database
a year ago
how have you been able to confirm that this issue is not with hetzner's network?
a year ago
unfortunately that doesn't confirm that this isn't an issue with hetzner's network
a year ago
does this VPS have a static IP?
I tried connecting from that VPS to another db in another railway environment
same issue
a year ago
are you able to renew that static IP?
a year ago
do you have a separate VPS on hetzner that you can try mysql from?
if yes, we need to move off railway
we have critical infrastructure that we cant swap out like this in production
a year ago
at this time we don't have definitive proof of who's blocking who
a year ago
may I ask why not move the services on the VPS to railway?
the services on vps needs multiple tcp ports, which railways fails to provide
a year ago
you would need more than 1 TCP port per app?
a year ago
ah so really it's neither gcp or hetzner's fault
a year ago
not the list gcp uses apparently
a year ago
I would go with bart's proposed solution
but i'd like to find a more permanent solution since we have an identical setup for production
a year ago
if gcp is blocking hetzner, there really nothing railway can do about that, they are at gcp's whim here
a year ago
railway isn't blocking your singular static IP, so we can rule railway out of the equation
a year ago
unless you tried to open well over 10k concurrent requests? this isn't a block from the railway side of things
a year ago
so does a singular app of yours need to open more than one TCP port?
there are 2 apps which needs us to run vps, one needs a tpc port with a custom subdomain on our own and another with access to like 4 ports
signup btw… could prob get brody@ https://unn.sh
a year ago
may I ask if you both are pro on railway? you seen to be missing the badges if so
a year ago
now you have them
10 months ago
Google blocks a littany of IP addresses due to Hertzner propensity to host...very sketchy shitDDOS: https://managingwp.io/live-blog/google-cloud-and-hetzner-connectivity-issues-gridpane-specifics/Violating Iranian Sanctions: https://blog.cloud66.com/hetzner-connectivity-issues-due-to-sanction-busting-activitiesMore: https://github.com/kubernetes/kops/issues/16466