Uploaded image for project: 'Red Hat Data Grid'
  1. Red Hat Data Grid
  2. JDG-5031

Scaling up 8.3 cluster with payload is slower compared to 8.2

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Critical Critical
    • RHDG 8.3 CD
    • RHDG 8.3 CD
    • Openshift Images
    • None
    • False
    • False
    • Hide

      Create a cluster:

      kind: Infinispan
apiVersion: infinispan.org/v1
metadata: 
  name: infinispan
  namespace: datagrid-suite-upgrade
spec: 
  expose: 
    type: Route
  replicas: 3
  security: 
    endpointAuthentication: false
    endpointEncryption: 
      type: None
  container: 
    cpu: 2000m
  service: 
    type: DataGrid

      Once up create persisted cache:

      {
  "distributed-cache": {
    "mode":"SYNC", 
    "encoding": {
      "media-type": "application/json"
    },
    "persistence": {
       "file-store":{}
    },
    "statistics":true
  }
}

      Upload data (ideally from within the container to speed it up):

      for x in {1..5000}
do
        curl -X PUT -d "{\"value\":\"${x}\"}" -H "Content-Type: text/plain" http://infinispan:11222/rest/v2/caches/testcache/key-${x}
done

      Scale cluster down to 1 and then back to 3. You'll notice second node is taking time to boot.

      Show
      Create a cluster: kind: Infinispan apiVersion: infinispan.org/v 1 metadata: name: infinispan namespace: datagrid-suite-upgrade spec: expose: type: Route replicas: 3 security: endpointAuthentication: false endpointEncryption: type: None container: cpu: 2000 m service: type: DataGrid Once up create persisted cache: { "distributed-cache" : { "mode" : "SYNC" , "encoding" : { "media-type" : "application/json" }, "persistence" : { "file-store" :{} }, "statistics" : true } } Upload data (ideally from within the container to speed it up): for x in {1..5000} do curl -X PUT -d "{\" value\ ":\" ${x}\ "}" -H "Content-Type: text/plain" http: //infinispan:11222/ rest /v2/caches/testcache/key-${x} done Scale cluster down to 1 and then back to 3. You'll notice second node is taking time to boot.

      If there cache with data exist and user scales up a cluster then the time to bring up another node is way higher compared to 8.2 and may even result in Initial state transfer timeout.

      While 8.2 clusters with 5000 persisted json entries is brough up in constant time, time required by 8.3 is at least 3 times longer potentially timing out.

        1. infinispan-1-infinispan.log
          5 kB
        2. infinispan-0-infinispan.log
          22 kB
        3. 1014523.txt
          178 kB
        4. log-restart-full
          3.83 MB
        5. log-0-full
          11.84 MB
        6. log-restared-8.2
          3.38 MB
        7. log-0-8.2
          8.65 MB

            wburns@redhat.com Will Burns
            pdrobek@redhat.com Pavel Drobek
            Pavel Drobek Pavel Drobek
            Votes:
            0 Vote for this issue
            Watchers:
            7 Start watching this issue

              Created:
              Updated:
              Resolved: