Event Platform/EventStreams/Administration
See EventStreams for an overview of the EventStreams service.
EventStreams is a service-template-node based service. It glues together KafkaSSE with common Wikimedia service features, like logging, error reporting, metrics, configuration and deployment.
Internally, EventStreams is available at eventstreams.svc.${::site}.wmnet. It is routed to by varnish and LVS from stream.wikimedia.org.
EventStreams in production is configured and deployed using WMFs Deployment pipeline.
Configuration
EventStreams is configured in the operations/deployment-charts repository. Configuration is spread between the defaults in charts/eventstreams and the production specific configuration in helmfile.d.
helmfile.d values.yaml files contain the list of allowed_streams that EventStreams will expose. The actual Kafka topics are retrieved from EventStreamConfig. Our event topics are prefixed by datacenter name. This is abstracted for EventStreams consumers via this mapping.
Kafka
EventStreams is backed by the main Kafka clusters. As of 2018-08, EventStreams is multi-DC capable. EventStreams in eqiad consumes from the Kafka main-eqiad cluster, and EventStreams in codfw consumes from the Kafka main-codfw cluster. Kafka MirrorMaker is responsible for mirroring the topics from eqiad to codfw and vice versa.
NodeJS Kafka Client
KafkaSSE uses node-rdkafka (as do other production NodeJS services that use Kafka).
Repositories
| Repository | Description |
|---|---|
| KafkaSSE (github) | Generic Kafka Consumer -> SSE NodeJS library. |
| eventstreams (github) | EventStreams implementation using KafkaSSE and service-template-node. |
| operations/deployment-charts | Helm chart repository for all production Kubernetes based services, including EventStreams. |
Deployment
See: Deployments on kubernetes
Submitting changes
Change to KafkaSSE library
KafkaSSE is hosted in Github, so you must either submit a pull request or push a change there.
kafka-sse is an npm dependency of EventStreams.
If you update kafka-sse, you should bump the package version and publish to npm: https://www.npmjs.com/package/kafka-sse
Change to mediawiki/services/eventstreams repository
EventStreams is hosted in gerrit. Use git review to submit patches. If you've modified the KafkaSSE repository, you should update the kafka-sse dependency version in package.json. Merged changes in this repository will result in a new Docker image being built.
To test locally:
# Build the new Docker image
DOCKER_BUILDKIT=1 docker build --target production -f .pipeline/blubber.yaml --platform=linux/amd64 . -t es
# Run the Docker image locally. The UI will be available at http://localhost:8092/
docker run --rm -it -v $(pwd):/es -w /es --net=host es
You can also test on deployment-prep (go in horizon.wikimedia.org -> deployment-prep -> filter for "event"), the Docker image to set is stated in the VM's puppet configuration. Once updated, run puppet on the Event Streams Beta VM and check that everything is fine (you can also use docker logs etc.. if you want).
Finally, the result will be displayed in https://stream-beta.wmflabs.org/
If you wan to see an event displayed for mediawiki.revision-create, you can edit https://en.wikipedia.beta.wmflabs.org/wiki/Polar_bear
For details on how to access and work with deployment-prep refer to Event Platform/Beta Deployments.
Update operations/deployment-charts repository
Once a new Docker image has been made, you'll need to update the image_version in helmfile.d eventstreams values.yaml.
Deploy
See: Deployments_on_kubernetes#Code_deployment/configuration_changes and Event Platform/Beta Deployments
Logs
Logs are sent to logstash. You can view them in Kibana.
Metrics
https://grafana.wikimedia.org/d/znIuUcsWz/eventstreams
Throughput limits
As of 2019-07, The public EventStreams stream.wikimedia.org endpoint is configured in varnish to only allow for 25 concurrent connections per varnish backend. There are 10 text varnishes in codfw and 8 in eqiad, so the varnish concurrent connection limit for EventStreams is 200 in eqiad and 250 in codfw for a total of 450 concurrent connections. We have had incidents where a rogue client spawns too many connections. EventStreams code has some primitive logic to try to reduce the number of concurrent connections from the same X-Client-IP, but this will not fully prevent the issue from happening. Check the total number of connections in https://grafana.wikimedia.org/dashboard/db/eventstreams if new connections receive a 502 error from varnish.
Alerts
EventStreams is configured with a monitoring check that will check that the /v2/stream/recentchange URL has data on it. This check is done to the public stream.wikimedia.org endpoint. If this public check fails, then likely all backend service processes have the same issue.