The client's observability is lacking. We use Sentry for error tracking but do not have client RUM and performance monitoring instrumentations. It means that we do not understand how real users use our client application in terms of performance/interactions. It limits our ability to assess and prioritize areas for improvement. We are forced to do it based on limited customer development sessions, anecdotal evidence, or one-time testing without the ability to monitor trends and proactively jump on the most impactful issues. Instrumenting our client application can reveal user issues that couldn't be observed or diagnosed any other way.
Measure of success
[ ] We have a robust process for client performance observability on the client, and it's documented in the handbook.
[ ] Relevant services and tools are configured to enable the client performance observability.
[ ] Auto instrumentations for client performance monitoring are configured.
[ ] The collected amount of app telemetry data is sufficient for running queries, visualizing traces, and asking questions about specific user interactions.
Solution Summary
[x] Evaluate the relative advantages and disadvantages of available performance monitoring tooling and determine the best path forward
[ ] Enable continuous web application bundlesize monitoring.
[x] #39736
[x] #39735
[x] #39737
[ ] #39738
[ ] Basic OpenTelemetry integration to enable gathering performance metrics immediately so that we can identify bottlenecks across the application and make targeted improvements that improve core workflow.
[x] #39273
[ ] #39276
[x] #39274
[ ] Integrate client OpenTelemetry data with backend OpenTelemetry collector
Problem to solve
The client's observability is lacking. We use Sentry for error tracking but do not have client RUM and performance monitoring instrumentations. It means that we do not understand how real users use our client application in terms of performance/interactions. It limits our ability to assess and prioritize areas for improvement. We are forced to do it based on limited customer development sessions, anecdotal evidence, or one-time testing without the ability to monitor trends and proactively jump on the most impactful issues. Instrumenting our client application can reveal user issues that couldn't be observed or diagnosed any other way.
Measure of success
Solution Summary
Artifacts:
What specific customers are we iterating on the problem and solution with?
Internal Sourcegraph developers
Delivery plan
Tracked issues
@valerybugakov
Completed
Legend