[bitnami/rabbitmq]: cpu use much higher after updating from 11.4.0 to 11.13.0

[oprudkyi]

Name and Version

bitnami/rabbitmq 11.13.0

What architecture are you using?

arm64

What steps will reproduce the bug?

• gke
• standalone (1 replica) and ha config (3 replicas, clustering = true)
• low load (up to 5-50 messages per second)
• update from 11.4.0 to 11.13.0 (also checked 11.12.1)
• check cpu load after few hours after update/restart

kubectl top pods -A | grep rabbit
rabbitmq                  rabbitmq-standalone-0                                            961m         169Mi

Are you using any custom parameters or values?

  values = [
    yamlencode({
      replicaCount          = 1
      podAntiAffinityPreset = "hard"
      auth = {
        username     = "..."
        password     = "..."
        erlangCookie = "..."
      }
      memoryHighWatermark = {
        enabled = true
        # https://www.rabbitmq.com/memory.html
        type  = "relative"
        value = "0.5"
      }
      clustering = {
        enabled   = false
        rebalance = false
      }
      maxAvailableSchedulers = "2"
      onlineSchedulers       = "2"
      resources = {
        requests = {
          memory = ""512Mi"
          cpu    = "200m"
        }
        limits = {
          memory = "2Gi"
          cpu    = "1"
        }
      }
      terminationGracePeriodSeconds = 15 //default 120, for preemptible <= 25 is supported
      service = {
        externalTrafficPolicy = "Local"
      }
      persistence = {
        enabled      = true
        storageClass = "premium-rwo"
        size         = "2Gi"
      }
      extraConfiguration = join("\n", [
        "default_vhost = some-name",
        "log.file.level = error",
        "log.console.level = error",
        "consumer_timeout = 1800000,
      ])
      nodeSelector = {
        "kubernetes.io/os" = "linux"
        role               = "common-services"
        preemptible        = "false"
      }
      metrics = {
        enabled = true
      }
      # don't allow to kill it by kube-dns
      priorityClassName = "system-cluster-critical"
    })
  ]

What is the expected behavior?

low cpu use

kubectl top pods -A | grep rabbit
rabbitmq                  rabbitmq-standalone-0                                            148m         239Mi

What do you see instead?

higher cpu use (~1CPU per process)

kubectl top pods -A | grep rabbit
rabbitmq                  rabbitmq-standalone-0                                            961m         169Mi

Additional information

it's weird, but just after update cpu usage is low - 70-200ms, but in and hour or few cpu usage raises up to ~1 cpu without any logs/errors etc
here metrics from 3 nodes ha cluster - the cpu started low and then nodes one by one started to use more cpu

also, on some clusters one replica still use low cpu for days while others start to use higher cpu

kubectl top pods -A | grep rabbitmq
rabbitmq                  rabbitmq-cluster-0                                               1019m        320Mi           
rabbitmq                  rabbitmq-cluster-1                                               103m         278Mi           
rabbitmq                  rabbitmq-cluster-2                                               1045m        324Mi   

I also tried to increase cpu limits, but without success

[oprudkyi]

also I tried to increase request/limit for mem, but without success.
rollback down to 11.4.0 helps though

[oprudkyi]

correction - rollback down to 11.4.0 didn't help . Either there are some saved changes on fs level or something like

[javsalgar]

Hi,

Thank you for opening the ticket. I believe this is a duplicate of #11116. Feel free to reopen it if you believe it is something different.