Orchestrator Resource Settings are not applied to MySQL CDC Sync

[SteZe85]

Helm Chart Version

2.2.0

What step the error happened?

None

Relevant information

This issue is connected to #68162

We have a CDC pipeline using the mysql-source connector which sometimes silently fails on orchestrator side with:
Terminating due to java.lang.OutOfMemoryError: Java heap space

We upgraded to Airbyte 2.2.0, which we understood to contain a fix for this problem, but the behaviour persists.

Configuration
The following values are set in the ConfigMap:

REPLICATION_ORCHESTRATOR_MEMORY_LIMIT: 4Gi
REPLICATION_ORCHESTRATOR_MEMORY_REQUEST: 2Gi

Expected behaviour

The orchestrator pod is created with the configured memory limit of 4Gi and request of 2Gi.

Actual behaviour

The sync job log still reports the default values, so the configured limits appear to be ignored:

platform INFO [orchestrator] image: airbyte/container-orchestrator:2.2.0 resources: ResourceRequirements(claims=[], limits={memory=1024Mi, cpu=1}, requests={memory=1024Mi, cpu=1}, additionalProperties={})

As a result, the OutOfMemoryError continues to occur intermittently.

Relevant log output

2026-08-17 09:00:03 platform INFO Building socket environment: transport=SOCKET, serialization=PROTOBUF
2026-08-17 09:00:03 platform INFO APPLY Stage: LAUNCH — (workloadId=bc1b2258-45ab-4c33-a9e3-e639c31ab45f_199457_0_sync)
2026-08-17 09:00:03 platform INFO [initContainer] image: airbyte/workload-init-container:2.2.0 resources: ResourceRequirements(claims=[], limits={memory=1024Mi, cpu=1}, requests={memory=1024Mi, cpu=1}, additionalProperties={})
2026-08-17 09:00:03 platform INFO Launching replication pod: replication-job-199457-attempt-0 (selectors = {}) with containers:
2026-08-17 09:00:03 platform INFO [source] image: airbyte/source-mysql:3.53.3 resources: ResourceRequirements(claims=[], limits={memory=8Gi, cpu=2}, requests={memory=2Gi, cpu=1}, additionalProperties={})
2026-08-17 09:00:03 platform INFO [destination] image: airbyte/destination-clickhouse:2.1.27 resources: ResourceRequirements(claims=[], limits={memory=8Gi, cpu=2}, requests={memory=2Gi, cpu=1}, additionalProperties={})
2026-08-17 09:00:03 platform INFO [orchestrator] image: airbyte/container-orchestrator:2.2.0 resources: ResourceRequirements(claims=[], limits={memory=1024Mi, cpu=1}, requests={memory=1024Mi, cpu=1}, additionalProperties={})

---

Internal Tracking: https://github.com/airbytehq/oncall/issues/11158

[octavia-bot]

AI Triage Starting...

Thank you for opening this issue! I'm an AI assistant that will do an initial review and may:

• Ask clarifying questions if needed
• Suggest relevant documentation or workarounds
• Flag this for internal visibility

Note: Community issues are reviewed on a best-effort basis and do not have a guaranteed response time. For additional help:

• Join the Airbyte Community Slack to connect with other users and community members
• If you're an Airbyte Cloud customer, you can open a support ticket or contact your account representative, referencing this issue URL

View playbook

• 🔗 Session URL: https://app.devin.ai/sessions/ed7f98ba296e44c687662208eeadf89b

✅ Devin AI session created successfully!

[airbyte-support-bot]

Hi SteZe85, thank you for reporting this and for including the launch log line and your ConfigMap values — that's exactly what we need.

We're looking into this. A related report is already being tracked internally, and we've linked your report to it so the team has the 2.2.0 data point.

A couple of things that would help the team narrow this down:

• Which values did you set in your Helm values file (as opposed to the raw ConfigMap)? Specifically whether you tried global.workloads.resources.containerOrchestrator.memory.request / .limit.
• Whether the environment variables are present in the workload-launcher pod itself (kubectl exec into it and check, or inspect its rendered deployment), since that's the component that builds the orchestrator pod spec.
• Whether the connection or actor has any resource requirements configured in the UI, which may interact with the global settings.

Please mask or remove any sensitive information (API keys, passwords, tokens, connection strings, hostnames) before sharing any further logs or manifests.

This issue is being tracked internally: https://github.com/airbytehq/oncall/issues/11158

---

Need more help? Join Airbyte Community Slack for peer support, or if you're a Cloud customer, open a support ticket referencing this URL.

[SteZe85]

These are my settings:

# kubectl -n airbyte-abctl get cm -o yaml | grep -i ORCHESTRATOR_MEMORY
    REPLICATION_ORCHESTRATOR_MEMORY_LIMIT: 4Gi
    REPLICATION_ORCHESTRATOR_MEMORY_REQUEST: 2Gi

# kubectl -n airbyte-abctl set env deploy/airbyte-abctl-workload-launcher --list | grep -i ORCHESTRATOR
CONTAINER_ORCHESTRATOR_MEMORY_REQUEST=2Gi
CONTAINER_ORCHESTRATOR_MEMORY_LIMIT=4Gi

But inside the running launcher pod, the orchestrator still runs with default:

# kubectl -n airbyte-abctl get pod replication-job-199696-attempt-0 \
  -o jsonpath='{range .spec.containers[*]}{.name}{"\t"}{.resources.limits.memory}{"\n"}{end}'> 
orchestrator    1Gi
source  4Gi
destination     4Gi