Install/Upgrade notes for GetVisibility Essentials chart

4.5.145

Redis volume

Rolled back Redis volume size tom 8 GB. If tenant had 4.5.144 installed with 16 GB, k3s can block upgrade with message:

cannot patch "gv-essentials-redis-master" with kind StatefulSet: StatefulSet.apps "gv-essentials-redis-master" is invalid: spec: Forbidden: updates to statefulset spec for fields other than 'replicas', 'ordinals', 'template', 'updateStrategy', 'persistentVolumeClaimRetentionPolicy' and 'minReadySeconds' are forbidden

If that is the case, delete Redis PVC redis-data-gv-essentials-redis-master-0 and retry the upgrade.

4.5.130

CRD PostgreSQL

Could be that there is an error while upgrading to that version.

Please then install Support Tools (gv-debug) at least in version 0.2.32 and make sure you checked:

Open

Expected output would be then

Then re-install gv-essentails

4.5.119

FIX in gv-debug

Elastic search CRD under helm control. If cluster does not get updates from fleet follow this guide

CRD elastic issue root cause

In the past we used to install CRD for elastic manually, now we want them being managed by the helm chart eck-operator 3.0.0 · elastic/elastic

However they are missing some labels, we added a job to the client-config from where fleet should push it to the cluster. Sometimes clusters are paused and not receiving fleet updates. Then when you try to upgrade to latest gv-essential you will have an error

Fix - relabel CRDs

To label CRDs please install gv-support-tool chart.

1. Make sure to check customize helm chart

   Open

   

2. On default page make sure you enabled

Open

3. Install and check logs

Pod which runs the script will be named like

logs should look like that

That should be it, then you can upgrade gv-essential

4.2.1

Defined CPU consumption limit for the following services

• Kafka

• Elasticsearch

PR reference: https://github.com/GetVisibility/rancher-charts/pull/136

4.0.1

FIXES AND ENHANCEMENTS

• Elasticsearch configuration UI improvements and bugfixes

4.0.0

ADDED

• Enabled Elasticsearch for all clusters
  Elasticsearch is planned to replace Postgres as the main data store for all products (including Synergy). We’re working to provide some degree of automation to update many clusters at once, but feel free to use latest 4.x version for manual upgrades if needed.

KNOWN ISSUES

Elasticsearch kind missing

When upgrading gv-essentials installed manually with Helm to version 4.x you may get the error Error: UPGRADE FAILED: unable to recognize "": no matches for kind "Elasticsearch"... .
This is because Helm cannot determine whether upgrading a CRD will break other things in the cluster, therefore, Helm v3 by design can create CRDs but is not able not upgrade them.

To address this issue install the CRDs manually:

In airgapped deployments, you can upload the elastic.yaml and run the following command before upgrading the chart:

Not enough memory

Wheen upgrading gv-essentials on a machine with 32 GB RAM via helm (without Rancher), Elasticsearch can get misconfigured to use 20 GB RAM, and the cluster will run out of memory. In this case, please amke sure you adjust the resource variables manually:

Decreasing storage size is not supported: an attempt was made to decrease storage size for claim elasticsearch-data

A volume for ES data was already created in this cluster, and it’s larger than entered size. Current size can be checked in Storage -> Persistent Volumes -> elasticsearch-data-gv-es-default-0 or via kubectl get pv , but usually it's 160 GB.

For helm installations, use --set eck-operator.settings.disk=160 --set eck-operator.settings.storage=160

3.2.0

FIXES AND ENHANCEMENTS

• Configured SaaS mode

3.1.2

FIXES AND ENHANCEMENTS

• Actually disable heap dumps on OOM in Elasticsearch

3.1.1

FIXES AND ENHANCEMENTS

• DIsabled heap dumps on OOM in Elasticsearch

3.1.0

FIXES AND ENHANCEMENTS

• Upgrade Zalando Postgres Operator to version 1.10.0

• Make chart compatible with Kubernetes 1.25.x

KNOWN ISSUES

When upgrading gv-essentials to version 3.1.x you’ll get the error Error: UPGRADE FAILED: error validating "": error validating data... .
This is because Helm cannot determine whether upgrading a CRD will break other things in the cluster, therefore, Helm v3 by design can create CRDs but is not able not upgrade them.

To address this issue run the following command before upgrading the chart:

Or you can download the file from the internet, save it locally and run the command kubectl apply -f crd_operatorconfigurations_v1_10_0.yml from within the cluster.