Build per-tenant ClickHouse write-routing for ingest (Tantivy still deferred)

ingest tags every record with a tenant_id Kafka header (built previously),
but nothing consumed it to actually route the write. This closes that for
ClickHouse: enterprise/cmd/enterprise-ingest (a second binary, mirroring
enterprise-api) reuses ingest/consumer's own flush loop unchanged, with
enterprise/internal/chwriter.Registry -- a per-tenant clickhousewriter.Writer
registry -- swapped in as the writer. A batch pulled from the single shared
Redpanda topic can mix records from many tenants, so WriteBatch groups by
TenantID and dispatches each group to its own tenant's connection, fail-
closed on an empty or unrecognized tenant_id.

ingest/consumer and ingest/clickhousewriter move out of internal/ (same
reason api/internal/* moved earlier this phase: enterprise/ can't import
anything under another module's internal/). Their New() constructors now
take small local Config structs instead of ingest/internal/config types,
so enterprise/ doesn't need that import either.

Building this surfaced a real bug: tenantprovision.ProvisionClickHouse
only granted SELECT on a tenant's ClickHouse user, correct for chrunner's
read-only use but not enough for chwriter reusing the same credential to
write -- every real per-tenant write would have failed closed with a
permission error. Fixed by widening the grant to SELECT, INSERT; no
cross-tenant boundary is crossed by also allowing INSERT within a
tenant's own database.

Helm gates enterprise-ingest's Deployment on the same
ingest.requireTenantCredential flag that already gates tag validation --
write-routing is meaningless without tagging already being required, so
they're one decision, not two. docker-compose.yml's version is a
disclosed, weaker approximation: it can't achieve Helm's genuine
-mode=server/-mode=consumer split, so with the enterprise profile active
both ingest and enterprise-ingest independently consume every message
via different consumer groups -- harmless duplication for local
verification only.

Not built: Tantivy's independent Redpanda consumer (search/src/consumer.rs)
still doesn't read the tenant_id header at all -- every record still lands
in the one shared index regardless of tenant. Not run: the live-ClickHouse-
gated tests (chwriter's cross-tenant routing test, tenantprovision's INSERT
regression test) -- no Docker/database access in this environment; they're
correct Go that has never executed, disclosed as such in docs/security/
threat-model.md and docs/phase-4-runbook.md §14.
This commit is contained in:
2026-08-14 19:26:09 -07:00
parent 17fdc212c2
commit 1de77b969f
26 changed files with 1355 additions and 267 deletions
+65 -17
View File
@@ -182,18 +182,14 @@ silently left out:
a cross-origin `fetch` with credentials from `web`'s origin needs
it), neither of which is verifiable in this environment without a
live backend and a browser session to exercise.
- **Ingest write-routing, for either storage engine** -- identity is now
real (see "Ingest tenant identity" below), but nothing consumes it
yet: `chrunner`/`searchclient` prove read isolation given tenant-
scoped data exists, and every record `ingest` produces is now tagged
with a real tenant ID, but neither `ingest/internal/consumer` (the
ClickHouse writer) nor `search/src/consumer.rs` (a completely
independent Redpanda consumer) reads that tag back to route the write
anywhere per-tenant. Every record still lands in the single shared
ClickHouse database and the single shared Tantivy index regardless of
tenant. A newly-provisioned tenant's storage is real and isolated, and
permanently empty. Now scoped, disclosed remaining work, not an
undesigned gap -- see `/docs/security/threat-model.md`.
- **Ingest write-routing, for Tantivy** -- `search/src/consumer.rs` (a
completely independent Redpanda consumer, not called through `ingest`
or `enterprise-ingest` at all) still writes every record into the one
shared (default) Tantivy index, regardless of tenant. The ClickHouse
half is now built (see "Ingest write-routing" below); Tantivy's is
real, disclosed, separate follow-up work -- a different codebase
(Rust) and a different consumer process, not just "the same fix
applied twice."
Deployment-topology routing (does traffic actually reach `enterprise-api`
instead of `api`) is no longer deferred -- both `deploy/helm/sentry` and
@@ -226,17 +222,67 @@ import (`ingest` is AGPL core), same "network boundary, not import
boundary" shape `api/authz.HTTPAuthorizer` already uses for the query
path.
**What this does not do**: change where a record is actually written.
See "Deliberately deferred" above -- attaching a verified tenant
identity as early as possible (right where the credential is presented)
was built as a self-contained first step; per-tenant write-routing for
both storage engines is separate, scoped follow-up work.
## Ingest write-routing (ClickHouse)
`enterprise/cmd/enterprise-ingest` is `ingest -mode=consumer`'s multi-
tenant alternative -- same "second binary" shape as `enterprise-api`
next to `api/cmd/api` (AGPL core must never import `enterprise/`, so the
tenant-aware wiring has to live in a binary that imports *into* core,
not the reverse). It reuses `ingest/consumer.Consumer`'s exact flush
loop unchanged, swapping in `enterprise/internal/chwriter.Registry` --
chrunner's write-side counterpart -- as the writer: one fully separate
`*ingest/clickhousewriter.Writer` (and the `driver.Conn` under it) per
tenant, built once at startup from `rbacstore.
ListProvisionedDataSources` (the same source of truth `chrunner` already
uses for reads). `WriteBatch` groups a Kafka batch's records by their
`tenant_id` tag and writes each tenant's group through its own
dedicated connection, refusing the whole call (matching `ingest/
consumer`'s existing all-or-nothing batch contract -- no offsets commit,
the batch redelivers) if any record is untagged or tagged with a tenant
that isn't provisioned. `ingest/consumer` and `ingest/clickhousewriter`
moved out of `internal/` for this -- same Go compiler-enforced
visibility reasoning as every other package this phase moved out of
`internal/` for a cross-module import (see `ingest/README.md`'s "Multi-
tenant write-routing" section).
A real bug was found and fixed while wiring this up:
`tenantprovision.ProvisionClickHouse` originally granted a tenant's
ClickHouse user `SELECT` only -- correct for `chrunner`'s query path,
but it would have made every real per-tenant write from `chwriter` fail
with a permission error, since it's the *same* credential used for
both. Fixed by granting `SELECT, INSERT` (not a second, separate
write-only credential -- there's no cross-tenant boundary crossed by
also granting INSERT within a tenant's own database, so one credential
for both directions is the simpler, still-correctly-scoped choice).
A real multi-tenant deployment runs `ingest -mode=server` (agent-facing,
tags records, unchanged) alongside `enterprise-ingest` (consumer,
per-tenant writes) *instead of* `ingest -mode=consumer` -- see `deploy/
helm/sentry`'s `ingest.requireTenantCredential` value (gates both the
credential-validation requirement and this mode split together, since
write-routing is only meaningful once records actually carry a
tenant_id to route on) and `docker-compose.yml`'s `enterprise-ingest`
service (a simpler opt-in there -- true `-mode=server`/`-mode=consumer`
exclusivity isn't wired in compose, a disclosed local-dev-only gap; see
that service's own comment).
Verified: `enterprise/internal/chwriter`'s fail-closed paths (empty/
unknown `tenant_id`) run genuinely without Docker (constructing a
`Registry` directly, bypassing `New`, which is the only part that would
dial ClickHouse); the actual per-tenant write-isolation probe
(`TestRegistryWritesEachTenantToItsOwnDatabase`) and the
`tenantprovision` INSERT-grant regression test are real integration
tests against a live ClickHouse, same `CHWRITER_TEST_CLICKHOUSE_ADDR`/
`TENANTPROVISION_TEST_CLICKHOUSE_ADDR` convention as every other
ClickHouse-backed test this phase -- not run against a live database in
this environment.
## Package layout
```
cmd/enterprise-auth/ config loading, OIDC discovery at startup, health/authorize/features/authorize-ingest endpoints, -mint-service-token, -create-tenant, -grant-membership-*, -revoke-membership-*, -list-memberships-tenant, -create-ingest-credential-tenant, -list-ingest-credentials-tenant, -revoke-ingest-credential
cmd/enterprise-api/ multi-tenant-aware alternative to api/cmd/api -- see its own doc comment
cmd/enterprise-ingest/ multi-tenant-aware alternative to ingest -mode=consumer -- see its own doc comment
internal/tenant/ the ID type -- see its package doc comment before touching it
internal/oidc/ coreos/go-oidc wiring: discovery, login redirect, code exchange + ID token verification
internal/saml/ crewjam/saml wiring: SP setup, login redirect, response parsing/validation
@@ -247,10 +293,12 @@ internal/rbacstore/ users/tenants/tenant_memberships/data_sources/dashb
internal/tenantprovision/ real ClickHouse CREATE DATABASE/USER/GRANT
internal/tenantcrd/ syncs -provision-tenant's real result into deploy/operator's Tenant CRD (K8s dynamic client, no cluster needed to test)
internal/chrunner/ tenant-scoped api/querylang/executor.SQLRunner
internal/chwriter/ tenant-scoped ingest/consumer.chWriter -- chrunner's write-side counterpart
internal/searchclient/ tenant-scoped api/querylang/executor.SearchClient
internal/audit/ append-only, hash-chained query audit log, plus the
api/queryapi.AuditLogger adapter (queryapi_adapter.go)
internal/apiconfig/ enterprise-api's own env-var config
internal/ingestconfig/ enterprise-ingest's own env-var config
internal/config/ enterprise-auth's env-var config
```