Skip to content

Latest commit

 

History

History
39 lines (30 loc) · 2.02 KB

README.md

File metadata and controls

39 lines (30 loc) · 2.02 KB

myblog-search

Application for searching my blog articles at https://blog.seanlee.site/

There are 147 articles: 106 articles in Chinese and 41 articles in English. All English articles have their counterpart in Chinese. There are pairs of English and Chinese articles for the same topic. The search bar provided by blogger.com does not recognize the above relationship, so there are duplication in the search results when a search keyword exists in both languages. This application is to avoid the duplication in search results.

The search application was based on Vespa Text Search Tutorial at https://docs.vespa.ai/en/tutorials/text-search.html but rewired by Elasticsearch (https://www.elastic.co/elasticsearch/) It is deployed on Google Kubernetes Engine at https://search.seanlee.site/?query=%E9%AD%9A

The back end is a single-node Elasticsearch server. It is deployed as a stateful set with persistent volumes for storing search index.

The middleware is a stateless Golang program to append parameters for Elasticsearch/Vespa to return search results in JSON format.

The frond end is a stateless reverse proxy by NGINX. It forwards queries to the middleware and render search results by Vue.js.

The last component of this search application is a cralwer that downloads the blog articles in ATOM format, convert the articles into Elasticsearch/Vespa document format in json format by Golang, and then feed the Elasticsearch/Vespa documents into the backend. It is deployed as a Kubernetes CronJob with a static persisent volume to retain the download blog feed. The retained feed is used for requesting only the recent updated blog feed instead of full feed. Also, the retained feed can be used for rebuilding/refeeding the search index.

The following is the data flow of this search application:

HTTP client --> NGINX --> Middleware --> Vespa <-- Crawler <-- Blog

Docker Compose is also used, but only in local development environment. It is for practicing and comparing the functional difference between Kubernetes and Docker Compose.