Problemsby GlyphKubernetes is too complex for most teams
13 posts from 13 people11d activeSteady, +1 this week
Posts per day
The posts behind it
Dead Link for Bluesky on PyPi Page
### Under which category would you file this issue? Airflow Core ### Apache Airflow version 3.3.2 ### What happened and how to reproduce it? The PyPi page for apache-airflow has a dead link (it links to which Bluesky can't resolve). ### What you think should happen instead? I...
Arrow provider: AdbcHook get_records forces transactions for autocommit-only ADBC drivers
...https://github.com/apache/airflow/pull/75988 explains the external RedisAdbcHook workaround; it does not fix the generic provider behavior. This report concerns that separate runtime compatibility gap. SQLExecuteQueryOperator hook selection/provider registration is separate...
`BaseTrigger.on_kill` fails to complete if it takes more than 1 second
### Under which category would you file this issue? Airflow Core ### Apache Airflow version 3.3.2 ### What happened and how to reproduce it? #### The problem If a Trigger's `on_kill` method takes more than 1 second to run, it doesn't complete. This can leave external work...
[Bug]: Vertex Model Garden (vertex_ai/openai/...) streaming drops reasoning_content, extra tool calls and usage details
### Description Streaming a reasoning model on a Vertex AI Model Garden dedicated endpoint (`vertex_ai/openai/ `, SGLang backend, GLM-5.3) returns no `reasoning_content` deltas through LiteLLM. Non-streaming returns the reasoning. The same model behind `hosted_vllm/` streams...
[Bug]: A guardrail scan that never ran is logged as `success`
### Description When a guardrail's `apply_guardrail` returns without calling its provider, `guardrail_information` still records an entry with status `success`. Monitoring can't tell "checked and clean" apart from "not checked". This can for example be seen on a Bedrock...
QueuePool Limit Reached Error for executing the dagster job through launchpad
### What's the issue? When we try to execute a job manually using launchpad to update a specific dbt asset only. The launchpad is not open and it gives this error. It does not happen everytime, at some time it allows us to run. I can see similar issue was raised earlier...
SerializedDAG.partial_subset modifies the original DAG when it contains a mapped TaskGroup (ReferenceError with nested TaskGroups)
### Under which category would you file this issue? Airflow Core ### Apache Airflow version 3.3.2 ### What happened and how to reproduce it? In `SerializedDAG.partial_subset()`...
Dashboard auto-refresh dropdown ignores DASHBOARD_AUTO_REFRESH_INTERVALS (6.1.0 and master)
### Bug description Bug description Since 6.1.0, the dashboard auto-refresh interval options are hard-coded in the frontend, and DASHBOARD_AUTO_REFRESH_INTERVALS from superset_config.py is ignored. In previous version this setting limited the options shown in the refresh...
[Bug]: Terraform resource `litellm_mcp_server.auth_type` does not support `bearer_token` input, resulting in a 422 error
### Description Creating an MCP server in Terraform with the `litellm_mcp_server` resource and setting the `auth_type` field to `bearer`, which is supported by the resource, we get a 422 error in response after apply. However, if set the value to`bearer_token`, we get an error...
[Bug]: coordination Redis from REDIS_* is probed once at boot, so a startup race leaves budgets per-pod for the life of the pod
...avoids this, because that path attaches the client without a ping. That is a fine workaround, but the env fallback exists precisely for deployments that do not set the block, and a transient failure silently turns budget enforcement per-pod. Fix options, happy to send a PR for...
Rendered template formatting broken: new line every word
### Under which category would you file this issue? Airflow Core ### Apache Airflow version 3.3.1 ### What happened and how to reproduce it? Run any SQL task (we have `dbt` or/and plain Snowflake SQL for instance). Open `Rendered Templates`, copy any `SQL`, paste it anywhere in...
Plugins: preinstall auto-update unloads a bundled plugin before the install fails, leaving it unregistered
...same mechanism with a network-timeout trigger was reported for 13.0.1 in #123110. ### How do we reproduce it? 1. Run `docker.io/grafana/grafana:13.2.1` with a read-only root filesystem and a writable `/var/lib/grafana` (this is what grafana-operator and the Helm chart do by...
Advice on deploying a 5-microservice stack on a tight student budget
So I've been working on this microservices project ( repo ) for a few months. I have it fully running locally using kind , now I want to learn how to deploy it on AWS or GCP following best practices to get hands on experience and record demo videos for my resume.. Here's the...
Companies and products named
About this problem
Evidence
13 posts from 13 people in 6 places, about 1 a week over 58 days. Mostly on apache/airflow, BerriAI/litellm, r/devops. Tools named alongside: Kubernetes, GitHub, Docker, Apache Airflow.
Frustration Frustration 0 of 3· Seen on GitHub issues, Reddit
How it was grouped
Posts that state a pain and match the "k8s-complexity" rule. First post Sep 7, 2026, latest Nov 3, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
The same problem elsewhere
Apache Airflow is too complex to run well
34 posts from 32 people on GitHub issues.
Kubernetes is too complex to run well
8 posts from 8 people on GitHub issues, Stack Overflow.
Claude Code is too complex to run well
5 posts from 4 people on GitHub issues.
Airbyte is too complex to run well
4 posts from 4 people on GitHub issues.
Google Cloud is too complex to run well
2 posts from 2 people on Reddit. Unverified.
History