1e3b643e94
The goal of this PR is to make search blitz usable for customers who want to run it against their instance for continuous performance analysis. Specifically, this does two things: 1) Enables configuring the set of queries to run with the `SEARCH_BLITZ_QUERY_FILE` env var 2) Packages the image in our standard format so we can publish it |
||
|---|---|---|
| .. | ||
| attribution | ||
| scripts | ||
| .gitignore | ||
| api.go | ||
| attribution.graphql | ||
| BUILD.bazel | ||
| client.go | ||
| config_test.go | ||
| config.go | ||
| Dockerfile | ||
| image_test.yaml | ||
| main.go | ||
| prometheus.go | ||
| protocol.go | ||
| queries_dogfood.txt | ||
| queries.txt | ||
| query.go | ||
| README.md | ||
| search.graphql | ||
| stream_client.go | ||
Search-Blitz
The purpose of Search-Blitz is to provide a baseline for our search performance.
Search-Blitz calls the stream and GraphQL APIs for typical
queries in regular intervals. Sourcegraph recognizes the Search-Blitz's
User-Agent and sends metrics to Prometheus.
The dashboard is accessible on Grafana, section "Sentinel queries".
In addition to the dashboard that we ship with Sourcegraph, Search-Blitz is deployed with a dedicated instance of Prometheus and Grafana.
How to track a query
Add the query to queries.txt.
For attribution search add a .txt file to the attribution directory.
A custom set of queries can be provided at runtime by pointing to a query file with the SEARCH_BLITZ_QUERY_FILE env var.
How to deploy
-
Merge your changes to main
-
Build and upload a new docker image:
./scripts/build.sh <next-version, e.g. 0.0.2> -
Update the image tag in deploy-sourcegraph-cloud
-
(Optional) Apply the new manifest
kubectl apply -f ./configure/search-blitz
How to access Search-Blitz's dedicated Grafana
kubectl port-forward search-blitz-0 3000:3000 -n monitoring