Signals: how to recognize deployment context
When you're reviewing a doc, looking at a screenshot, reading a ticket, or auditing search results, you often need to know which deployment type the content is about without anyone telling you explicitly.
This page collects the cues that signal a piece of content is rooted in a specific deployment type. Use them to:
- Spot when an
applies_totag is missing or wrong - Catch content that's quietly scoped to one deployment type without saying so
- Quickly orient yourself in unfamiliar areas of the docs
| Signal | Self-managed | ECK | ECE | ECH | Serverless |
|---|---|---|---|---|---|
| URL pattern in screenshots or links | Direct host URL (for example, https://localhost:5601) |
(none — Kubernetes service URL) | Customer-defined admin URL (for example, admin.<host>) |
cloud.elastic.co |
cloud.elastic.co/projects |
| Terminology for the unit of work | Cluster | Cluster (Kubernetes-native: Custom Resource) | Deployment | Deployment | Project |
| UI or surface named | None (config on disk) | Kubernetes CRDs and manifests | ECE Cloud UI | Elastic Cloud Console | Project settings |
| Files or mechanisms mentioned | elasticsearch.yml, kibana.yml, keystore |
Kubernetes volumes, secrets, init containers | Bundles, user settings, plugins | User settings, bundles, plugins | (no user-accessible config files) |
| Features mentioned | All Elasticsearch APIs, ILM, plugins | All + operator-managed features | Same as ECH | All except some ECE-only restrictions; supports plugins, bundles, private links, agentless | Limited subset: no ILM, no plugins, no user-controlled snapshots, no BYOK |
| Telltale phrase | "edit the file," "SSH into the node" | "apply the manifest," "patch the CR" | "in the ECE Cloud UI," "as a platform admin" | "in Elastic Cloud Console" | "in your project," "in project settings" |
When you encounter a doc or PR:
- Scan for signals from the table above.
- Check whether the page-level
applies_tomatches the signals you see. - If there's a mismatch:
- If the page is about a deployment-agnostic concept that mentions one deployment's surface — flag the surface mention, not the page scope
- If the page is genuinely scoped to a specific deployment but isn't tagged that way — fix the
applies_to
For the full review checklist, refer to Review checklist.
| What you see | What it signals | What to do if the page should be cross-deployment |
|---|---|---|
| "Log in to Elastic Cloud Console and open Kibana" on a generic Kibana feature page | Elastic Cloud Hosted (orchestrator surface) | Replace with deployment-agnostic phrasing (for example, "Open Kibana") and link to a single generic "access Kibana" doc |
"Edit elasticsearch.yml to set this value" on a setting-configuration page |
Self-managed direct file access. Other deployment types use user settings (ECH, ECE), CRDs (ECK), or don't expose the setting (Serverless) | Use an applies-switch per surface, or link to Stack settings |
A code block starting with apiVersion: elasticsearch.k8s.elastic.co/v1 |
ECK manifest | Either scope the surrounding content to ECK or clearly mark the code block as one of several deployment-specific approaches |
- Documenting deployment types — primer and content classification
- Review checklist — symptoms to look for during PR review
- Strategies for deployment-type variation — what to do when content varies