There are several reports of OOMs and of jobs “hanging” or pods being killed, but in the provided sources I do
not see a clear, documented, 2.0‑specific bug where: 1. the
source container is OOM‑killed, 2. the
destination keeps processing “legitimately”,
and 3. the
sync is finally marked as successful. What we do see: - Kubernetes / self‑managed cases where a pod is killed (OOM or node change) and the job then appears to hang or stay “running” rather than being properly failed, which was traced to missing/enforced timeouts or heartbeat handling in the platform. [[Sync attempt hangs](
https://github.com/airbytehq/airbyte/issues/48879)] - Multiple OOM issues on specific source connectors (MySQL, CockroachDB, DynamoDB, Stripe, Pub/Sub + HubSpot, CDC sources, etc.), but these are described as jobs failing with OOM or getting repeatedly OOM‑killed, not as successfully completed syncs. [[CockroachDB OOM](
https://discuss.airbyte.io/t/1261); [source-mysql OOM](
https://github.com/airbytehq/airbyte/issues/51014); [DynamoDB OOM fix](
https://github.com/airbytehq/airbyte/issues/61600); [Stripe memory leak](
https://github.com/airbytehq/airbyte/issues/60813); [HubSpot→Pub/Sub OOM](
https://discuss.airbyte.io/t/889); [Java heapspace CDC](
https://github.com/airbytehq/airbyte/issues/51616)] - A Jira case where a destination container fails but the pod keeps running and logs suggest processing continues even after a STOPPED state, which is closer to a “job state vs. reality” mismatch, but still not exactly “source OOM, destination runs, sync marked successful.” [[source-jira big project](
https://github.com/airbytehq/airbyte/issues/60917)] None of these explicitly state that in
Airbyte 2.0 the platform will mark a sync as
successful when the source has been OOM‑killed while the destination proceeds. So, based on the supplied material: - There
are known OOM and lifecycle/state‑handling issues (particularly with specific connectors and some k8s/job lifecycle edge cases). - The knowledge sources do
not explicitly document a 2.0‑specific, generally acknowledged bug matching exactly “source OOM‑killed + destination continues + sync marked successful.” If you’re seeing that exact behavior on 2.x, the evidence here isn’t enough to confirm it as a known tracked issue; opening a new GitHub issue with logs and version details would be the next step.