Problems with OpenSearch & ComponentPack - v8

HCL Connections Connections Forum
,
1

Hi,

I am trying to setup OpenSearch pods. I started with "Opensearch-Master" after initializing connections-env, volumes, etc...I am using the following command to install:

helm upgrade opensearch-master v-connections-helm/opensearch -i --version 1.3.7-20230117-205850 --namespace connections -f opensearch_master.yml

opensearch_master.yml:

image:
  repository:               hclcr.io/cnx

clusterName:                opensearch-cluster

nodeGroup:                  “master”

masterService:              “opensearch-cluster-master”

pemkeyPass:                 locarno2023

imagePullSecrets:


    

  • name: myregkey
    • 



roles:


    

  • master
    • 



service:

labels:                   {}

labelsHeadless:           {}

type:                     NodePort

nodePort:                 30099

nodePortTransport:        “”


createSecret:               false


resources:

limits:

cpu:                    “1”

memory:                 “1024Mi”

requests:

cpu:                    “0.1”

memory:                 “768Mi”


replicas:                   3


"kubectl logs" does not work because pod is still not initialized "Init:CrashLoopBackOff".
"kubectl events -A" : 

connections   9m11s       Normal    Created            pod/opensearch-cluster-master-2                     Created container fsgroup-volume
connections   9m11s       Normal    Started            pod/opensearch-cluster-master-2                     Started container fsgroup-volume
connections   8m29s       Normal    Pulling            pod/opensearch-cluster-master-2                     Pulling image "busybox:latest"
connections   8m28s       Normal    Pulled             pod/opensearch-cluster-master-2                     Successfully pulled image "busybox:latest" in 902.247168ms (902.270348ms including waiting)
connections   8m23s       Normal    Pulling            pod/opensearch-cluster-master-0                     Pulling image "busybox:latest"
connections   8m18s       Normal    Pulling            pod/opensearch-cluster-master-1                     Pulling image "busybox:latest"
connections   5m2s        Warning   BackOff            pod/opensearch-cluster-master-0                     Back-off restarting failed container
connections   4m57s       Warning   BackOff            pod/opensearch-cluster-master-2                     Back-off restarting failed container
connections   4m45s       Warning   BackOff            pod/opensearch-cluster-master-1                     Back-off restarting failed container

ES5 & ES7 worked smoothly on this Kubernetes-Host(all-in-one deployment, only master). Since I have much less parameters to provide with CP8 & everything is in an online-repo I was wondering: What have I done wrong ? Why is the OS-master-pod not starting ?

Any help/ideas appreciated.
I am already in contact with HCL but no progress - I hope the community can provide some additional input.

Many thanks in advance !
Kind regards,
Jan

2

Hi.

If you have already checked this, disregard:

https://foxutech.medium.com/kubernetes-crashloopbackoff-how-to-troubleshoot-940dbb16bc84

"If you get the back-off restarting failed container message this means that you are dealing with a temporary resource overload, as a result of an activity spike."

regards,

Ariel

3

Hi

Have you checked your pods with 'kubectl describe' ?

4

We were able to sort-out all Opensearch-related problems.
The big points I remember that finally solved the issue at the end:

1. sysctl -w vm.max_map_count=262144 needed to be executed. The pod log indicated not enough ressources for max_map_count are available
2. I chosed a custom name for the opensearch-cluster. I named it "opensearchcluster" but it MUST be "opensearch-cluster". eventhough .j2-template files & documentation indicate you can choose any value it has to be the string "opensearch-cluster"