Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Ankur Pata
    @ankurpata
    Btw, I am not using traefik right now. I do not have a domain name. I am just trying to run it on a cloud with an ip.
    Loan Laux
    @loan-laux
    For everything. You’d put each service of the stack behind a reverse proxy.
    And particularily Hydra + the IDP. The rest you can perhaps skip, but Hydra will complain if it doesn’t have a real URL with HTTPS.
    Ankur Pata
    @ankurpata
    Can I skip the authentication and run Admin without it? I think graphql queries wont work without them.
    Loan Laux
    @loan-laux
    You can, but you won’t be able to log in. Kinda defeats the purpose.
    Javier Daza
    @javierdaza
    @loan-laux ok. My tbs-backend security group allows port 80 and 443. The one for my EC2ContainerServiceCluster is allowing ports 80, 443, 4444, 4100, 4080, 4000 and 3000
    Ankur Pata
    @ankurpata
    Right, still How can I directly access the admin ?
    Loan Laux
    @loan-laux
    @javierdaza Should be good then. Anything in your container logs?
    @ankurpata You can’t if you don’t have an OAuth server set up, like Hydra. But you can get a free .tk domain or something like this, right?
    Ankur Pata
    @ankurpata
    Yes, I can look for a domain. But, without domain we cant configure routes on a cloud server I guess then right ?
    Loan Laux
    @loan-laux
    Not sue what you have in mind @ankurpata
    Ankur Pata
    @ankurpata
    lol. Just want to know if a domain is mandatory to deploy reactioncommerce on cloud server :)
    If not. Is there any documentation explaining to do so
    Javier Daza
    @javierdaza
    @ankurpata As far as I've seen all the examples need a domain. I think you should get one free domain to avoid the Redirect URL is using an insecure protocol error. Also have in mind that he told you Hydra needs a real URL with HTTPS
    Ankur Pata
    @ankurpata
    right. I guess yes. I'll try with a secure domain. Thanks
    Javier Daza
    @javierdaza
    @loan-laux These is what I found in my logs
    INFO Reaction:  Initializing MongoDB replica set...
    INFO Reaction:  Connecting to MongoDB...
    INFO Reaction:  Connected to MongoDB. Database name: reaction
    
    WARN[39m Reaction: [36m Failed to initialize a MongoDB replica set. This may result in errors or some things not working. Error: Error initializing replica set: not authorized on admin to execute command { replSetInitiate: { _id: "123", version: 3, protocolVersion: 1, members: [ { _id: 0, host: "cluster.something.mongodb.net", priority: 100 } ] }, lsid: { id: UUID("12345678") }, $clusterTime: { clusterTime: Timestamp(1590779116, 1), signature: { hash: BinData(0, 12345ABCDE), keyId: 123456789 } }, $db: "admin" } (Unauthorized)
    Ankur Pata
    @ankurpata
    I am getting this error:
    Cannot find module /usr/local/src/app/plugins.json imported from /usr/local/src/app/node_modules/@reactioncommerce/api-core/src/importPluginsJSONFile.js
    While trying to deploy from the ansible playbook. I changed readction's repos to my custom repos by creating new config.local.mk
    Has anyone facing this problem before ?
    Ankur Pata
    @ankurpata
    @loan-laux Hi ,
    I added my plugins in docker-compose-dev.yml before and it was working.
    I am guessing its not picking from "-dev" in deploying on production
    Ankur Pata
    @ankurpata
    Whats. the right way of deploying api plugins on production ?
    Loan Laux
    @loan-laux
    @ankurpata publish them to NPM (publicly or privately), and npm install them.
    @javierdaza Can you try and pass the env variable REACTION_SHOULD_INIT_REPLICA_SET=false to your reaction API container?
    Ankur Pata
    @ankurpata
    Okay, Do I need to add it in the plugins.json ?
    Loan Laux
    @loan-laux
    Affirmative @ankurpata
    Ankur Pata
    @ankurpata
    Hi @loan-laux , I am still getting this error: :
    ERROR Reaction: Cannot find module /usr/local/src/app/plugins.json imported from /usr/local/src/app/node_modules/@reactioncommerce/api-core/src/importPluginsJSONFile.js
    Ankur Pata
    @ankurpata
    Updating to 3.7.1 fixed this issue.
    Ankur Pata
    @ankurpata
    But my custom plugin dint install I guess. I can not see the mutations and extensions from the new plugin.
    Loan Laux
    @loan-laux
    @ankurpata Is it in your package.json and your plugins.json?
    Javier Daza
    @javierdaza
    Thanks @loan-laux the REACTION_SHOULD_INIT_REPLICA_SET=false env variable solved the issue.
    Now I have to check why the task is failing the ELB health checks to get rid of the 503 Service Temporarily Unavailable error
    Javier Daza
    @javierdaza
    I don't understand that at some point the API looks like is waiting ok GraphQL subscriptions ready at wss://api.mydomain.com/graphql (port 3000) and then a couple of minutes later it kills the task SIGTERM.
    The target group protocol is HTTP, path is /graphql, port is 3000 and success codes is 400.
    Is there anything wrong with my target group configuration or health check?
    Javier Daza
    @javierdaza
    Solved. It was an issue with the availability zones (different subnets)
    Loan Laux
    @loan-laux
    Glad to hear that @javierdaza!
    Javier Daza
    @javierdaza
    @loan-laux About the tutorial. For deploying the Hydra's public API, do you still recommend using image oryd/hydra:v1.0.8 or a newer version can do the job also?
    Loan Laux
    @loan-laux
    @javierdaza if there are no breaking changes, then yes by all means!
    Javier Daza
    @javierdaza
    @loan-laux I am using oryd/hydra:v1.1.1
    I checked the containers statuses and it seems that my hydra container is running but my hydra-migrate is stopped after 3 SQL migrations. Is it normal to be stopped?
    Loan Laux
    @loan-laux
    @javierdaza if it exited with status 0, then everything is fine. It means that it succeeded. If not, check the logs and you should get an indication of what’s wrong.
    Javier Daza
    @javierdaza
    @loan-laux thanks. It was fine. A couple of things I realized of the tutorial so far(RC 3):
    1. In the section Creating your cluster, the content is duplicated
    2. For my tbs-hydra-admin task to work I had to enable port 4445 in the security groups of the EC2ContainerService-<your-cluster-name> instance
    Loan Laux
    @loan-laux
    Thank you so much for the feedback, @javierdaza! I’ll look into this ASAP. Is your environment running as it should now?
    Javier Daza
    @javierdaza
    @loan-laux I am currently configuring the tbs-admin task. I haven't finished it yet
    Javier Daza
    @javierdaza
    Hi
    Have you ever received an error net::ERR_CERT_COMMON_NAME_INVALID when trying to deploy tbs-admin??
    From firefox I get a CORS related warning
    Cross origin request blocked: The same origin policy does not allow remote resources to be read
    It is https://hydra.mydomain.com/.well-known/openid-configuration. (Reason: CORS request unsuccessful)
    Javier Daza
    @javierdaza
    @loan-laux I tried using this env variable SERVE_ADMIN_CORS_ENABLED = true in my hydra containers but it didn't work. Do you happen to had issues with CORS before?
    Javier Daza
    @javierdaza
    On the other hand, Chrome is telling me a different story
    DEBUG [react-context-oidc] : overriddenConfiguration 
    {client_id: "reaction-admin", redirect_uri: "https://admin.mydomain.com/authentication/callback", response_type: "code", post_logout_redirect_uri: "https://admin.mydomain.com", scope: "openid", …}
    authority: "https://hydra.mydomain.com"
    automaticSilentRenew: true
    client_id: "reaction-admin"
    loadUserInfo: true
    post_logout_redirect_uri: "https://admin.mydomain.com"
    redirect_uri: "https://admin.mydomain.com/authentication/callback"
    response_type: "code"
    scope: "openid"
    silent_redirect_uri: "https://admin.mydomain.com/authentication/silent_callback"
    triggerAuthFlow: true
    __proto__: Object
    INFO [react-context-oidc] : Protection :  true
    INFO [react-context-oidc] : Protected component mounted
    [Violation] 'setInterval' handler took 228ms
    Javier Daza
    @javierdaza
    Looks like the problem is with
    https://hydra.mydomain.com/.well-known/openid-configuration
    net::ERR_CONNECTION_TIMED_OUT
    Thomas
    @lacrx

    I'm having trouble loading the admin on a prod environment of the reaction admin. I've noticed in the websocket messages that I'm able to authenticate, but then I get this response:

    a["{\"msg\":\"updated\",\"methods\":[\"3\"]}"]
    a["{\"msg\":\"result\",\"id\":\"3\",\"error\":{\"isClientSafe\":true,\"error\":400,\"reason\":\"Match failed\",\"message\":\"Match failed [400]\",\"errorType\":\"Meteor.Error\"}}"]

    and then shows a blank admin area with only the Reaction logo and a logout buttun. From what I understand, "match failed" is an unhandled Meteor error. Has anyone experienced this before? Why might that be happening?

    Thomas
    @lacrx
    and I should add that I also am able to register a new user on the backend. In the messages, it shows the new user being created and signing in, then throws that same "match failed" error, then redirects me to a backend/undefined url
    Thomas
    @lacrx
    oh also i'm using v2.0.0
    Thomas
    @lacrx

    i also see this warning in the backend browser console:

    name: "reaction-client",
    hostname: "backend-prod.examplestore.com",
    pid: undefined,
    level: 40,
    msg: "Missing storefront home URL. Please set this from the shop settings panel so that customer users can be redirected to your storefront.",
    time: Date Sat Jun 06 2020 17:43:42 GMT-0700 (Pacific Daylight Time),
    v: 0

    i wonder if the pid being undefined is an issue

    Loan Laux
    @loan-laux
    @lacrx Any errors from the backend container?
    Loan Laux
    @loan-laux
    @javierdaza I had this issue once but I don’t remember how I handled it… Sorry, I know that’s not useful to you. Can you manually reach https://hydra.mydomain.com/.well-known/openid-configuration?