September 2018 Summaries

This tutorial guides you through setting up an observable system for a GraphQL application using Hasura GraphQL Engine and Honeycomb. An observable system allows you to understand, optimize, and control your GraphQL server by capturing important metrics such as query execution time, response status codes, and user context. By integrating with Honeycomb, you can create visual views of logs based on filters, set up alerts for anomalies, and gain a holistic view of your GraphQL queries to identify performance issues or errors. The setup includes deploying Hasura GraphQL Engine on Google Kubernetes Engine (GKE), setting up the Honeycomb Agent on Kubernetes, and sending GraphQL Engine logs to Honeycomb. Finally, you can create dashboards with filters for granular details and set up alerts for anomalies in query execution time or other metrics.

Gatsby is a popular framework for building statically rendered React sites. However, in some cases, data needed for these sites may come from databases like Postgres. To address this issue, the gatsby-source-graphql plugin can be used to set any GraphQL API as a Gatsby data source. Additionally, Hasura can expose an existing Postgres database over GraphQL instantly. A boilerplate and tutorial by Karthik Venkateswaran are available for quick setup and integration of large datasets from Postgres into Gatsby sites using GraphQL.

Hacktoberfest 2018 is a month-long celebration of open source software that takes place every October. This year, Hasura will be participating in the event and encourages developers to contribute to their GraphQL Engine repository for a chance to receive some cool Hasura swag. The Hasura GraphQL Engine is an instant, realtime GraphQL API server over Postgres with webhook triggers on database events for asynchronous business logic. Issues across all components of the engine are available for taking up and have been marked as hacktoberfest. Participants can submit a pull request to the Hasura GraphQL Engine repository between October 1st and 31st, and those who contribute will receive rewards such as sticker packs and t-shirts.

This guide explains how to set up uptime monitoring for Hasura GraphQL Engine on Google Kubernetes Engine (GKE) using DataDog. The process involves setting up Hasura GraphQL Engine on GKE, installing the Datadog agent on Kubernetes, and configuring fine-grained metrics with kube-state-metrics. Once these steps are completed, users can create monitoring alerts in DataDog for uptime monitoring of their applications running on a production Kubernetes cluster.

At Hasura, Cypress test automation was used to test their GraphQL-engine's UI. However, running all tests synchronously took more than 30 minutes, slowing down the development process. To improve this, parallel testing was introduced by spawning multiple processes in a single container. Despite facing issues with Xvfb and inconsistent results, Cypress v3.1.0 allowed for parallelism on CI systems using the --parallel flag. By enabling parallel tests and recording them through Cypress Dashboard service, Hasura reduced their overall build times from ~45 minutes to ~15 minutes.

Hasura has released a public preview of Event Triggers on its open-source GraphQL engine. This feature allows developers to trigger webhooks on database events in just 60 seconds. It is designed for serverless deployments and can be used to create push notifications, purge CDN or cache, update search indexes, and trigger asynchronous business logic. Event Triggers are atomic, reliable, horizontally scalable, and offer debugging and observability features. The combination of real-time GraphQL and serverless functions is particularly exciting for developers.