D
Dokploy•6mo ago
max.d

Zero downtime for NextJS

Hey guys I'm having difficulty setting up 0 downtime for a NextJS app with 4 Replicas. This is my config: Health Check 
{
  "Test": [
    "CMD",
    "curl",
    "-f",
    "http://localhost:3000"
  ],
  "Interval": 30000000000,
  "Timeout": 10000000000,
  "StartPeriod": 30000000000,
  "Retries": 3
}
{
  "Test": [
    "CMD",
    "curl",
    "-f",
    "http://localhost:3000"
  ],
  "Interval": 30000000000,
  "Timeout": 10000000000,
  "StartPeriod": 30000000000,
  "Retries": 3
}
Update Config 
{
  "Parallelism": 1,
  "Delay": 10000000000,
  "FailureAction": "rollback",
  "Order": "start-first"
}
{
  "Parallelism": 1,
  "Delay": 10000000000,
  "FailureAction": "rollback",
  "Order": "start-first"
}
Additional Traefik config modification with sticky sessions 
  services:
    web-3d327d:
      loadBalancer:
        servers:
          - url: http://web-3d327d:3000
        passHostHeader: true
        sticky:
          cookie:
            httpOnly: true
  services:
    web-3d327d:
      loadBalancer:
        servers:
          - url: http://web-3d327d:3000
        passHostHeader: true
        sticky:
          cookie:
            httpOnly: true
After deploy NextJS serves new HTML but is unable to find new static files. I initially thought that this is due to sessions being balanced between old/new replicas incorrectly, but sticky sessions should've taken care of that. What am I missing?
2 Replies
𖤬𝔫𝔡𝔶
𖤬𝔫𝔡𝔶•2mo ago
^ bump, having the same issue @Siumauricio, after deploying the load balancer makes it so some users are loading the backend from an older container, we have this Traefik config: 
http:
  middlewares:
    # Rate-limit middleware
    rate-limit:
      rateLimit:
        average: XX  # Requests per second on average
        burst: XX    # Maximum requests allowed to burst
        period: Xs   # Time period for the average calculation
        sourceCriterion:
          ipStrategy:
            depth: 0  # Use the client IP directly
    
    # Large upload middleware
    large-upload:
      buffering:
        maxRequestBodyBytes: XXXXXXXXXX  # Size in bytes
        memRequestBodyBytes: XXXXXXXXX   # Threshold before using disk
          
  routers:
    [service-name]-router-XXXX:
      rule: Host(`example.com`)
      service: [service-name]-service-XXXX
      middlewares:
        - redirect-to-https
      entryPoints:
        - web
        
    [service-name]-router-websecure-XXXX:
      rule: Host(`example.com`)
      service: [service-name]-service-XXXX
      middlewares: 
        - rate-limit
        - large-upload
      entryPoints:
        - websecure
      tls:
        certResolver: letsencrypt
        
  services:
    [service-name]-service-XXXX:
      loadBalancer:
        servers:
          - url: http://[internal-service-name]:XXXX
        passHostHeader: true
        sticky:
          cookie:
            name: example_sticky
            secure: true
            httpOnly: true
            sameSite: "lax"
http:
  middlewares:
    # Rate-limit middleware
    rate-limit:
      rateLimit:
        average: XX  # Requests per second on average
        burst: XX    # Maximum requests allowed to burst
        period: Xs   # Time period for the average calculation
        sourceCriterion:
          ipStrategy:
            depth: 0  # Use the client IP directly
    
    # Large upload middleware
    large-upload:
      buffering:
        maxRequestBodyBytes: XXXXXXXXXX  # Size in bytes
        memRequestBodyBytes: XXXXXXXXX   # Threshold before using disk
          
  routers:
    [service-name]-router-XXXX:
      rule: Host(`example.com`)
      service: [service-name]-service-XXXX
      middlewares:
        - redirect-to-https
      entryPoints:
        - web
        
    [service-name]-router-websecure-XXXX:
      rule: Host(`example.com`)
      service: [service-name]-service-XXXX
      middlewares: 
        - rate-limit
        - large-upload
      entryPoints:
        - websecure
      tls:
        certResolver: letsencrypt
        
  services:
    [service-name]-service-XXXX:
      loadBalancer:
        servers:
          - url: http://[internal-service-name]:XXXX
        passHostHeader: true
        sticky:
          cookie:
            name: example_sticky
            secure: true
            httpOnly: true
            sameSite: "lax"
max.d
max.dOP•2mo ago
I've created a ticket here: https://github.com/Dokploy/dokploy/issues/1487 It looks like Traefik is not able to route requests properly. Cookie is set but requests are still routed via WRR.
GitHub
Unable to use sticky sessions in traefik · Issue #1487 · Dokploy/do...
To Reproduce I have an application running via Dockerfile. I have 4 replicas and application config modified as follows: services: development-experiments-z8rb1q-service-30: loadBalancer: sticky: c...

Did you find this page helpful?