The two primary pillars of the Cloud Team are Availability and Observability as defined in RFC 498
This team ensures that Sourcegraph.com has the same reliability and availability as other world-class SaaS offerings. This team is also responsible for Observability monitoring and tooling to ensure that we are meeting these goals.
More can be found in our Cloud Vision
The Cloud DevOps team is responsible for the infrastructure used to host Sourcegraph.com. This includes but is not limited to dashboard and observability, uptime and reliability, and managing our cloud provider resources. This team works closely with the other teams in the Cloud org to ensure sourcegraph.com is available and functional for our users. Notably, this team has the ability to slow or stop rollouts to Sourcegraph.com if needed to improve stability.
This team is responsible for
- Continuous deployment of Sourcegraph.com
- Cloud monitoring infrastructure (Prometheus / Grafana)
- Managed instances
- The best way to contact the
cloud-devopsteam is in the #cloud-devops slack channel. If it is urgent, please cc
@devops-supportin your message.
- Issues with
team/devopslabels and @team/cloud-devops team on GitHub
- Please do not
@cloud-devops-teamor directly message individual teammates - this is to try and protect their focus. Instead, use the
@devops-supporthandle, and the On-Call DevOps teammate will be notified.
- How to deploy a code change to the Cloud
- Large release (rollout release) process
- How to make configuration changes to sourcegraph.com
- Datadog monitoring
- How to add or modify DNS Records
- Disaster Recovery
- How to resize disks in StatefulSet
- How to use preprod aka staging
- Persistent disk backup schedule
- Silencing Alerts
- Cloud Postgres
- Managed instances operation
We primarily work within the following repositories:
- Our Cloud GitOps repo
- Managed Instance configuration repo
The DevOps GitHub Project board is the single source of truth.
We don’t have sprint or cycles. We are mostly task-driven under the quarterly goals.
- This is the good time to call out issues you would like to discuess with everyone
- Review our GitHub Project board to prioritize tasks for the week
- A review of what we did for learing purposes
We use geekbot to keep others informed of what’s going on asynchronously. This is a good time to share your progress and ask for help to remove any blockers.
We maintain an on-call rotation in Opsgenie. Reponsibilities of the teammate who is on-call include:
- Acknowledging incoming alerts
- Initiating incident procedures
- Publishing postmortems
- Adding issues to the DevOps GitHub board with the label
devops/supportfor external support requests, such as Slack messages addressed to
If the on-call teammate reaches the end of their shift with unresolved issues, they inform the rest of the team in #cloud-devops-internal.
- Assist the Cloud-SaaS team with RFC 525
- Stabilize our CD environment
- Create a pre-production environment for Cloud
- Standardize our monitoring, logging and error reporting systems
- Migrate our zonal cluster to a regional cluster and document the process
- Complete assigned security tasks on behalf of security and provide evidence of completion