Hi there,
Is it possible to restart a specific deployment with an API call?
This is to save memory and CPU and free those up when there is no volume of requests.
#Possible to restart a specific deployment via API?
oblique forge
rich bearBOT
queen reef
does the cron template not fit your usecase for this?
oblique forge
It does, but it sometimes restarts even when there is an ongoing job, and that job is impacted
want to restart when there is no ongoing job.
queen reef
gotcha, do you know golang?
oblique forge
not right now, but i think i might be familiar
queen reef
it's simple syntax, read the source code of the cron template, it's golang
queen reef
oblique forge
const serviceInstanceRedeployMutation = mutation serviceInstanceRedeploy($environmentId: String!, $serviceId: String!) { serviceInstanceRedeploy(environmentId: $environmentId, serviceId: $serviceId) };
const getLatestDeploymentQuery = query Deployments($first: Int, $input: DeploymentListInput!) { deployments(input: $input, first: $first) { edges { node { id status } } } };
could you please confirm if i'm running this correctly?
i've noticed i can only restart the service once
while i do get a 200, the services not really restart, is this something to do with deployment id?
async function redeployService(serviceId: string): Promise<{ serviceId: string; status: string; error?: string }> {
if (!RAILWAY_API_ENDPOINT) {
throw new Error('RAILWAY_BACKEND_URL is not defined in the environment variables');
}
try {
const latestDeploymentId = await getLatestDeploymentID(serviceId);
const response = await fetch(RAILWAY_API_ENDPOINT, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': `Bearer ${process.env.RAILWAY_API_TOKEN}`,
},
body: JSON.stringify({
query: serviceInstanceRedeployMutation,
variables: {
environmentId: process.env.RAILWAY_ENVIRONMENT_ID,
serviceId: serviceId,
deploymentId: latestDeploymentId,
},
}),
});
if (!response.ok) {
const errorText = await response.text();
throw new Error(`HTTP error! status: ${response.status}, body: ${errorText}`);
}
const data = await response.json();
if (data.errors) {
throw new Error(JSON.stringify(data.errors));
}
return { serviceId, status: 'redeployed' };
} catch (error: any) {
console.error(Error redeploying service ${serviceId}:, error);
return { serviceId, status: 'failed', error: error.message };
}
}
queen reef
you want to restart the service, not redeploy it
oblique forge
i can confirm i'm using correct details in the backend
ah okay, so this one?
mutation deploymentRestart($id: String!) {
deploymentRestart(id: $id)
}
queen reef
yep
oblique forge
sigh, thanks man, i'll try with this
queen reef
sounds good
maybe at some point it will be worth it to invest some time into your code so you dont have to keep restarting it to save memory?
oblique forge
Yes, i've tried doing multiple things, but it's apparently an issue with a package i'm using
Also, same issue, it restarts the first time i run the code after making changes,
const getLatestDeploymentQuery = `
query deployments($first: Int, $input: DeploymentListInput!) {
deployments(first: $first, input: $input) {
edges {
node {
id
status
createdAt
}
}
}
}
`;
const deploymentRestartMutation = `
mutation deploymentRestart($id: String!) {
deploymentRestart(id: $id)
}
`;
i tried restarting again, but services don't restart
no errors, getting 200
there a limit on how many times i can restart a service per hour or maybe minutes?
queen reef
no there is not
oblique forge
i guess there's something in the code then
i'll figure out, thanks for all the help!
queen reef
no problem!
oblique forge
correctly gets response for deployments to restart.
but deployments not restarted
works fine when i call graphql api via postman. restarts fine as well
queen reef
how are you sure you are getting the correct deployment Id
oblique forge
it would show an error if deployment ids are correct, i printed and manually verified these ids too
queen reef
I just have the strangest feeling you are running redeploys for a service in some other project or environment
oblique forge
i have had that feeling too, but it looks like the ids are correct
i'm printing as i run those and verified with the ones i have in my dashboard
queen reef
something is incorrect, otherwise it would work haha
oblique forge
next i'm fetching these dynamically
based on service ids
it would all fail if service ids were in correct
queen reef
think about this logically, if my cron template can restart your service, that narrows it down to it being a code issue on your side
oblique forge
i'm guessing some sort of caching? i can tell you it does 100% what it's supposed to when i call it via Postman
queen reef
wish I knew what the issue was though, sorry
railway doesn't cache API calls, this is a coding error as proven by my cron template working
oblique forge
;_;
queen reef
I'm sorry I don't have an answer for you but this is 100% a code issue
