ReadMeFeed / Community knowledge
Real questions. Useful conversations. Find the people who know your stack.
The goal is to silence repeated notifications for the same error that occurs across multiple service instances in a Traefik Mesh deployment, while still being alerted to genuine problems. This requires knowing how the mesh groups alerts before they reach Alertmanager, so that silencing rules can be applied without suppressing distinct incidents. However, the
Pub/Sub subscription retry policy controls minimum and maximum backoff for redelivery after nack or ack deadline expiry. The configuration affects delivery timing, not delivery semantics. The documented delivery guarantee is at-least-once. The API surfaces messageId per delivery, but there is no built-in deduplication window or correlation identifier provide
In a Stylus codebase with many partials that all depend on a shared library of mixins and variables, I need to pick a consistent inclusion strategy. Stylus documents two directives: @import inlines the target file on every reference, while @require includes a given file only once per compilation. The concrete constraint is stylesheet size: the shared file is
Goal: Ensure that retrying a failed HTTP request does not result in duplicate writes to the backend. Constraints: Elm’s update function is pure, so emitting the same Msg repeatedly produces the same Cmd; the Http module provides no built‑in retry, backoff, or deduplication mechanisms, and the runtime does not track in‑flight requests. Uncertainty: The core d
Goal Determine how to guarantee that audit events forwarded from Teleport to an external sink are not duplicated when the Teleport backend retries delivery after a transient failure. Constraints & Uncertainty Teleport’s audit sink delivery is documented as at‑least‑once – retries are automatic but no exactly‑once guarantee is enforced. The platform emits
We are designing a typed Akka service where a producer actor sends commands to a consumer that performs a database write. Since Akka messaging is at-most-once by default, any retry mechanism (supervision restart, explicit resend, or ReliableDelivery redelivery) can deliver the same logical command more than once, so the write must not be duplicated. The plan
traefikmesh
prometheus
google
stylus
akka