clickstack 3.0.0 defaults collector image tag to 2.19.0 while appVersion is 2.27.0

[mfroembgen]

The clickstack chart 3.0.0 reports appVersion: 2.27.0, but the default bundled ClickStack collector image still points at 2.19.0.

Repro:

helm show chart clickstack --repo https://clickhouse.github.io/ClickStack-helm-charts --version 3.0.0 | grep appVersion
helm show values clickstack --repo https://clickhouse.github.io/ClickStack-helm-charts --version 3.0.0 | sed -n '/otel-collector:/,/^[^[:space:]]/p'

Observed:

appVersion: 2.27.0

otel-collector:
  image:
    repository: docker.clickhouse.com/clickhouse/clickstack-otel-collector
    tag: "2.19.0"

Expected: the default collector image tag should probably align with the chart app version / current ClickStack collector release, i.e. 2.27.0, unless the older collector tag is intentional and should be documented.

[nijave]

looks like this is causing a crash for me

  'exporters' error reading configuration for "clickhouse": decoding failed due to the following error(s):
  'clickhouseexporter.Config' has invalid keys: json

[ChrsMark]

Ran into this on a fresh kind cluster today. Here's the full root-cause chain and confirmed workaround.

Root cause

The clickstack chart 3.0.0 ships appVersion: 2.27.0 but pins the otel-collector sub-chart image at tag: "2.19.0":

otel-collector:
  image:
    repository: docker.clickhouse.com/clickhouse/clickstack-otel-collector
    tag: "2.19.0"

The my-clickstack-app pod runs version 2.27.0 and pushes an OpAMP config to the collector that includes a json: key in the ClickHouse exporter:

exporters:
    clickhouse:
        json: ${env:HYPERDX_OTEL_EXPORTER_CLICKHOUSE_JSON_ENABLE:-false}
        ...

The 2.19.0 collector binary (otelcontribcol 0.145.0) does not recognise this key and exits immediately with:

'exporters' error reading configuration for "clickhouse": decoding failed due to the following error(s):
'clickhouseexporter.Config' has invalid keys: json

This surfaces as an endless CrashLoopBackOff on the my-clickstack-otel-collector pod. Because the supervisor catches the exit before the health-check endpoint comes up, the pod stays 0/1 and all downstream OTel collectors that forward to :4318 see connection refused.

Confirmed workaround

Override the image tag at install/upgrade time to match the app version:

helm install my-clickstack clickstack/clickstack \
  --set "otel-collector.image.tag=2.27.0" \
  ...

or for an existing release:

helm upgrade my-clickstack clickstack/clickstack \
  --set "otel-collector.image.tag=2.27.0" \
  --reuse-values

After the rollout the pod comes up cleanly and data flows through to ClickHouse.

Fix

The simplest fix is to align otel-collector.image.tag with appVersion in Chart.yaml — either pin it to 2.27.0 or derive it from the chart's own appVersion so the two stay in sync going forward.