feat(store): logical backup and isolated restore drill - #102
Merged
seonghobae merged 2 commits intoAug 23, 2026
Conversation
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
seonghobae
added a commit
that referenced
this pull request
Aug 23, 2026
Base automatically changed from
feat/issue-81-outbox-retention
to
feat/issue-80-postgres-rustls
August 23, 2026 18:02
Issue #80 remainder stacked on #101. GET /api/backup exports a hashed tenant snapshot; POST /api/backup restores after schema and payload-hash checks; POST /api/backup/drill restores into an isolated tenant, compares unmasked invariants, and drops the drill rows. Declared RPO is last successful export; declared RTO is 60s. File/memory adapters report backup=disabled. Do not re-implement rustls, outbox, or retention.
seonghobae
force-pushed
the
feat/issue-80-backup-restore
branch
from
August 23, 2026 18:04
0e8aa44 to
5e9e6aa
Compare
seonghobae
merged commit Aug 23, 2026
321e792
into
feat/issue-80-postgres-rustls
2 of 3 checks passed
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
Comment on lines
+1678
to
+1694
| async fn export_backup(client: &mut Client, tenant_id: &str) -> Result<ControlPlaneBackup, String> { | ||
| let snapshot = load_snapshot(client, tenant_id) | ||
| .await? | ||
| .ok_or_else(|| format!("tenant {tenant_id} has no snapshot to back up"))?; | ||
| let outbox = list_outbox(client, tenant_id, i64::MAX).await?; | ||
| let receipts = list_receipts(client, tenant_id).await?; | ||
| ControlPlaneBackup { | ||
| schema_version: MIGRATION_VERSION, | ||
| tenant_id: tenant_id.to_string(), | ||
| created_unix: unix_now_i64(), | ||
| snapshot, | ||
| outbox, | ||
| receipts, | ||
| payload_hash: String::new(), | ||
| } | ||
| .seal() | ||
| } |
There was a problem hiding this comment.
🟡 Backup export is not point-in-time consistent
export_backup reads the snapshot, the outbox (list_outbox), and the receipts (list_receipts) in three separate committed transactions. In PostgreSQL mode a leased outbox worker runs concurrently and every gateway event runs append_security_event, inserting a security_event and its outbox_message in one transaction. A write landing between the reads yields an artifact whose events, outbox rows, and receipts disagree, and restore reproduces that inconsistency.
Prompt for agents
export_backup in src/control_plane.rs runs load_snapshot, list_outbox, and list_receipts as three independent transactions, each of which sets tenant context and commits on its own. Because a leased outbox worker and gateway append_security_event calls mutate security_event/outbox_message/outbox_receipt concurrently in PostgreSQL mode, the exported ControlPlaneBackup can capture these three views at different points in time, producing an internally inconsistent artifact (e.g. an outbox_message without its security_event, or a receipt without its message). Consider performing the whole export within a single transaction (a single repeatable-read/serializable transaction that sets wardnet.tenant_id once and reads all tables) so the artifact is a true point-in-time snapshot, matching the one-transaction consistency guarantee used by save_snapshot.
Was this helpful? React with 👍 or 👎 to provide feedback.
Comment on lines
+499
to
+530
| pub async fn restore_drill(&self) -> Result<BackupDrillReport, String> { | ||
| let started = Instant::now(); | ||
| let backup = self.logical_backup().await?; | ||
| let isolated = format!("restore-drill-{}-{}", std::process::id(), unix_now_i64()); | ||
| let mut client = self.client.lock().await; | ||
| restore_backup(&mut client, &isolated, &backup).await?; | ||
| let restored = export_backup(&mut client, &isolated).await?; | ||
| drop_tenant(&mut client, &isolated).await?; | ||
| let source_hash = backup.semantic_hash()?; | ||
| let restored_hash = restored.semantic_hash()?; | ||
| let passed = source_hash == restored_hash | ||
| && restored.snapshot.routes == backup.snapshot.routes | ||
| && restored.snapshot.events == backup.snapshot.events | ||
| && restored.snapshot.threats == backup.snapshot.threats | ||
| && restored.snapshot.dnsbl == backup.snapshot.dnsbl | ||
| && restored.outbox.len() == backup.outbox.len() | ||
| && restored.receipts.len() == backup.receipts.len(); | ||
| let duration_ms = u64::try_from(started.elapsed().as_millis()).unwrap_or(u64::MAX); | ||
| Ok(BackupDrillReport { | ||
| passed, | ||
| duration_ms, | ||
| rpo: BACKUP_RPO.to_string(), | ||
| rto_budget_ms: BACKUP_RTO_BUDGET_MS, | ||
| source_hash, | ||
| restored_hash, | ||
| route_count: backup.snapshot.routes.len(), | ||
| event_count: backup.snapshot.events.len(), | ||
| outbox_count: backup.outbox.len(), | ||
| receipt_count: backup.receipts.len(), | ||
| isolated_tenant_id: isolated, | ||
| }) | ||
| } |
There was a problem hiding this comment.
🟡 Failed restore drill leaks the isolated tenant
restore_drill commits the restored rows into the isolated tenant, then re-exports and only afterward calls drop_tenant. The re-export uses ?, so any failure there returns early and skips drop_tenant, leaving the throwaway tenant's committed rows in every table. Repeated failed drills accumulate orphan data.
Prompt for agents
In restore_drill (src/control_plane.rs), the isolated tenant is populated by restore_backup (which commits its own transaction) and only dropped by drop_tenant after export_backup succeeds. If export_backup returns an error, the `?` propagates and drop_tenant is never called, leaking the committed drill-tenant rows. Ensure the isolated tenant is always cleaned up even on the error path — e.g. run the re-export and comparison, but wrap the post-restore steps so drop_tenant runs regardless of success/failure before returning the error (a guard/defer-style cleanup or an explicit cleanup on the error branch).
Was this helpful? React with 👍 or 👎 to provide feedback.
Comment on lines
+601
to
+629
| fn semantic_hash(&self) -> Result<String, String> { | ||
| let mut snapshot = self.snapshot.clone(); | ||
| snapshot.commercial.tenant_id.clear(); | ||
| let mut outbox: Vec<_> = self | ||
| .outbox | ||
| .iter() | ||
| .map(|message| { | ||
| ( | ||
| message.idempotency_key.clone(), | ||
| message.payload_hash.clone(), | ||
| message.message_status.clone(), | ||
| message.payload_json.clone(), | ||
| ) | ||
| }) | ||
| .collect(); | ||
| outbox.sort(); | ||
| let mut receipts: Vec<_> = self | ||
| .receipts | ||
| .iter() | ||
| .map(|row| (row.idempotency_key.clone(), row.receipt_evidence.clone())) | ||
| .collect(); | ||
| receipts.sort(); | ||
| let body = serde_json::json!({ | ||
| "snapshot": snapshot, | ||
| "outbox": outbox, | ||
| "receipts": receipts, | ||
| }) | ||
| .to_string(); | ||
| Ok(outbox::payload_hash(&body)) |
There was a problem hiding this comment.
📝 Info: semantic_hash / verify are deterministic
verify and the drill's passed check are stable: AppData has no map fields (only Vecs/scalars), semantic_hash sorts the outbox/receipt tuples and clears commercial.tenant_id, and those tuples omit tenant_id. A round-trip through an isolated tenant therefore hashes equal.
Was this helpful? React with 👍 or 👎 to provide feedback.
Comment on lines
+1760
to
+1762
| for message in &backup.outbox { | ||
| insert_restored_outbox(&tx, tenant_id, message).await?; | ||
| } |
There was a problem hiding this comment.
📝 Info: Restore re-arms pending outbox rows
restore_backup re-inserts outbox_message rows verbatim, including pending ones, so the worker re-dispatches them to stdout SIEM after a live restore even when receipts already exist. This matches the documented at-least-once contract, but operators should expect duplicate SIEM lines after a restore.
Was this helpful? React with 👍 or 👎 to provide feedback.
Comment on lines
+1268
to
+1279
| match plane.load().await { | ||
| Ok(Some(loaded)) => { | ||
| *state.inner.write().await = loaded; | ||
| } | ||
| Ok(None) => { | ||
| return error( | ||
| StatusCode::INTERNAL_SERVER_ERROR, | ||
| "restore committed but tenant snapshot is empty", | ||
| ); | ||
| } | ||
| Err(message) => return error(StatusCode::INTERNAL_SERVER_ERROR, message), | ||
| } |
There was a problem hiding this comment.
📝 Info: Live restore can diverge DB from memory
In the restore handler plane.restore_logical_backup commits the DB restore before plane.load(). If load() fails transiently the handler returns 500 while the DB is already restored and state.inner still holds pre-restore data, leaving them divergent until the next mutation or restart.
Was this helpful? React with 👍 or 👎 to provide feedback.
| pub async fn restore_drill(&self) -> Result<BackupDrillReport, String> { | ||
| let started = Instant::now(); | ||
| let backup = self.logical_backup().await?; | ||
| let isolated = format!("restore-drill-{}-{}", std::process::id(), unix_now_i64()); |
There was a problem hiding this comment.
📝 Info: Drill tenant id can collide within a second
restore_drill names the isolated tenant restore-drill-{pid}-{unix_now_i64()} at second resolution. Two drills in the same process within one second collide on that id; since the lock is released between logical_backup and the restore, concurrent drills can interleave and interfere on the shared tenant. Unlikely, but not guaranteed unique.
Was this helpful? React with 👍 or 👎 to provide feedback.
Comment on lines
+584
to
+599
| pub fn verify(&self) -> Result<(), String> { | ||
| if self.schema_version != MIGRATION_VERSION { | ||
| return Err(format!( | ||
| "backup schema_version {} is unsupported; expected {MIGRATION_VERSION}", | ||
| self.schema_version | ||
| )); | ||
| } | ||
| if self.tenant_id.trim().is_empty() { | ||
| return Err("backup tenant_id must be non-empty".to_string()); | ||
| } | ||
| let expected = outbox::payload_hash(&self.unsigned_json()?); | ||
| if expected != self.payload_hash { | ||
| return Err("backup payload_hash does not match contents".to_string()); | ||
| } | ||
| Ok(()) | ||
| } |
There was a problem hiding this comment.
🟨 Backup integrity check uses an unkeyed hash
ControlPlaneBackup::verify recomputes an unkeyed SHA-256 and compares it to the embedded payload_hash. The hash travels with the data and uses no secret, so a modified artifact can carry a freshly recomputed matching hash. The check detects corruption, not tampering. Restore is gated by an admin write token, so any bypass presupposes a privileged caller.
Was this helpful? React with 👍 or 👎 to provide feedback.
This was referenced Aug 23, 2026
seonghobae
added a commit
that referenced
this pull request
Aug 23, 2026
* feat(store): rustls for production PostgreSQL sslmode=require Issue #80 remainder. sslmode=require/verify-ca/verify-full connect with rustls and Mozilla roots; certificates are always verified. allow/prefer are still rejected so the process cannot silently drop to plaintext. Live test against plaintext CI postgres proves fail-closed. Do not re-implement the postgres gate or the outbox. * docs: record PR #100 in the product-technical gap baseline * fix(store): rewrite verify-full sslmode for tokio-postgres 0.7 Still-valid #100 Devin finding. tokio-postgres 0.7 only parses disable/prefer/require. Map verify-ca/verify-full to require before connect; rustls still verifies certificates. Password query-lookalikes are left untouched. * feat(store): bound outbox listing and prune processed rows (#101) * feat(store): bound outbox listing and prune processed rows Still-valid #99 finding. GET /api/outbox returns at most EVENT_LIMIT rows (dead letters, then pending, then leased, then processed). Processed outbox_message rows prune to that cap; receipts and dead letters stay. Do not re-implement the outbox, postgres gate, or rustls. * docs: record PR #101 in the product-technical gap baseline * fix(store): prune processed outbox to EVENT_LIMIT on save and ack Still-valid #101 Devin finding. Snapshot save and worker ack used LIST_LIMIT (1000) while append used operator EVENT_LIMIT. Store the configured cap on PostgresPlane so all three paths retain the same processed-row bound. Receipts and dead letters stay. * feat(store): logical backup and isolated restore drill (#102) * feat(store): logical backup and isolated restore drill Issue #80 remainder stacked on #101. GET /api/backup exports a hashed tenant snapshot; POST /api/backup restores after schema and payload-hash checks; POST /api/backup/drill restores into an isolated tenant, compares unmasked invariants, and drops the drill rows. Declared RPO is last successful export; declared RTO is 60s. File/memory adapters report backup=disabled. Do not re-implement rustls, outbox, or retention. * docs: record PR #102 in the product-technical gap baseline * feat(store): non-owner PostgreSQL runtime role after migrate (#103) * feat(store): non-owner PostgreSQL runtime role after migrate Still-valid #98 finding. CI connects as a superuser, which bypasses FORCE RLS. Migrations stay on the login role, then SET ROLE wardnet_runtime (NOSUPERUSER, NOBYPASSRLS, not table owner). Missing tenant GUC yields no rows; DROP TABLE and DISABLE RLS are denied. Do not re-implement rustls, outbox, retention, or backup/restore. * docs: record PR #103 in the product-technical gap baseline * fix(store): restore logical backups across role-only schema versions v3 only provisions wardnet_runtime and does not change table shape. verify() accepts schema 2 through the current migration version so a role-only upgrade cannot void the last pre-upgrade logical backup. * feat(store): HASH-partition security_event by tenant (#104) * feat(store): HASH-partition security_event by tenant Convert unpartitioned security_event to PARTITION BY HASH (tenant_id) with eight children under pg_advisory_lock. Rows keep unmasked client IPs and paths. /healthz.event_partitions reports the child count. Logical restore still accepts schema 2 through the current version. * docs: record PR #104 in the product-technical gap baseline
This was referenced Aug 23, 2026
seonghobae
added a commit
that referenced
this pull request
Aug 26, 2026
* feat(store): require PostgreSQL as the production control plane Non-loopback binds fail closed without CONTROL_PLANE_DATABASE_URL. Migrations create 3NF two-word tables with default-deny RLS. Snapshot persist commits policy rows and audit records in one transaction. Loopback still uses the JSON file or memory adapter. * feat(store): transactional outbox and leased workers Issue #81 first slice on the PostgreSQL control plane. Security events append with an outbox row in one transaction instead of rewriting the snapshot. Workers claim with SKIP LOCKED, retry, dead-letter, and record unique receipts. Stdout SIEM is at-least-once; the receipt is the exactly-once ack. Also deterministic ORDER BY on postgres loads (still-valid #98 finding). Do not re-implement the postgres gate. * feat(store): rustls for production PostgreSQL sslmode=require (#100) * feat(store): rustls for production PostgreSQL sslmode=require Issue #80 remainder. sslmode=require/verify-ca/verify-full connect with rustls and Mozilla roots; certificates are always verified. allow/prefer are still rejected so the process cannot silently drop to plaintext. Live test against plaintext CI postgres proves fail-closed. Do not re-implement the postgres gate or the outbox. * docs: record PR #100 in the product-technical gap baseline * fix(store): rewrite verify-full sslmode for tokio-postgres 0.7 Still-valid #100 Devin finding. tokio-postgres 0.7 only parses disable/prefer/require. Map verify-ca/verify-full to require before connect; rustls still verifies certificates. Password query-lookalikes are left untouched. * feat(store): bound outbox listing and prune processed rows (#101) * feat(store): bound outbox listing and prune processed rows Still-valid #99 finding. GET /api/outbox returns at most EVENT_LIMIT rows (dead letters, then pending, then leased, then processed). Processed outbox_message rows prune to that cap; receipts and dead letters stay. Do not re-implement the outbox, postgres gate, or rustls. * docs: record PR #101 in the product-technical gap baseline * fix(store): prune processed outbox to EVENT_LIMIT on save and ack Still-valid #101 Devin finding. Snapshot save and worker ack used LIST_LIMIT (1000) while append used operator EVENT_LIMIT. Store the configured cap on PostgresPlane so all three paths retain the same processed-row bound. Receipts and dead letters stay. * feat(store): logical backup and isolated restore drill (#102) * feat(store): logical backup and isolated restore drill Issue #80 remainder stacked on #101. GET /api/backup exports a hashed tenant snapshot; POST /api/backup restores after schema and payload-hash checks; POST /api/backup/drill restores into an isolated tenant, compares unmasked invariants, and drops the drill rows. Declared RPO is last successful export; declared RTO is 60s. File/memory adapters report backup=disabled. Do not re-implement rustls, outbox, or retention. * docs: record PR #102 in the product-technical gap baseline * feat(store): non-owner PostgreSQL runtime role after migrate (#103) * feat(store): non-owner PostgreSQL runtime role after migrate Still-valid #98 finding. CI connects as a superuser, which bypasses FORCE RLS. Migrations stay on the login role, then SET ROLE wardnet_runtime (NOSUPERUSER, NOBYPASSRLS, not table owner). Missing tenant GUC yields no rows; DROP TABLE and DISABLE RLS are denied. Do not re-implement rustls, outbox, retention, or backup/restore. * docs: record PR #103 in the product-technical gap baseline * fix(store): restore logical backups across role-only schema versions v3 only provisions wardnet_runtime and does not change table shape. verify() accepts schema 2 through the current migration version so a role-only upgrade cannot void the last pre-upgrade logical backup. * feat(store): HASH-partition security_event by tenant (#104) * feat(store): HASH-partition security_event by tenant Convert unpartitioned security_event to PARTITION BY HASH (tenant_id) with eight children under pg_advisory_lock. Rows keep unmasked client IPs and paths. /healthz.event_partitions reports the child count. Logical restore still accepts schema 2 through the current version. * docs: record PR #104 in the product-technical gap baseline
seonghobae
added a commit
that referenced
this pull request
Aug 26, 2026
* feat(security): fail-closed destination policy for outbound HTTP One DestinationPolicy mediates gateway upstreams, threat-intel fetches, Clearfolio, SOC LLM, and the Coraza sidecar URL. Private, loopback, link-local, CGNAT, and metadata classes are denied unless DESTINATION_ALLOWLIST (or loopback development) permits them; DESTINATION_DENYLIST wins. Clients ignore ambient HTTP proxies and do not follow redirects. Refs #79. * docs: record PR #96 in the product-technical gap baseline * feat(security): harden destination policy per-IP CIDR and readiness order CIDR allowlist matches apply per resolved address, authorize non-default ports, and reject prefixes outside the address-family width. IPv6 site-local is a denied class. Hostnames that merely contain 0x are not hex IP literals. AppState constructors default to production policy; seeded fixtures opt into development. Blocking DNS runs on spawn_blocking with a timeout. Persistence and destination-list validation complete before the readiness line. * feat(security): pin outbound HTTP to evaluated destination addresses After destination policy allows a host, the reqwest client resolves only those IPs so a rebinding answer cannot reach loopback, private, or metadata classes. Host and SNI stay on the original name. Unpinned hostnames fail closed instead of falling back to OS DNS. * feat(waf): evaluate live gateway transactions with in-process libcoraza Issue #86 remainder: dlopen operator-supplied libcoraza and drive the C ABI on each /gateway request. Missing library or empty ruleset fail closed before bind. CI stays hermetic with a fixture cdylib that exports the same symbols. * docs: record PR #97 in the product-technical gap baseline * feat(waf): forward bounded client headers into in-process libcoraza In-process transactions now receive the same forwarded-header allowlist as the sidecar path (host, user-agent, accept, content-type, referer, origin, x-requested-with, x-forwarded-for, x-real-ip, cookie — never Authorization; 32 headers / 8 KiB caps enforced by proven_engine::engine_forwarded_headers). Each header crosses the C ABI via coraza_add_request_header before process_request_headers, so CRS rules that inspect headers evaluate real client input instead of a synthetic Host only. Brings in the PR #95 sidecar hardening via merge so both engines share one allowlist implementation and one status/bound contract. Behavioral header-battery evidence lands with the issue-11 battery fixture (PR #110); this slice ships the plumbing and keeps the stub contract unchanged. * feat(store): require PostgreSQL as the production control plane (#98) * feat(store): require PostgreSQL as the production control plane Non-loopback binds fail closed without CONTROL_PLANE_DATABASE_URL. Migrations create 3NF two-word tables with default-deny RLS. Snapshot persist commits policy rows and audit records in one transaction. Loopback still uses the JSON file or memory adapter. * feat(store): transactional outbox and leased workers Issue #81 first slice on the PostgreSQL control plane. Security events append with an outbox row in one transaction instead of rewriting the snapshot. Workers claim with SKIP LOCKED, retry, dead-letter, and record unique receipts. Stdout SIEM is at-least-once; the receipt is the exactly-once ack. Also deterministic ORDER BY on postgres loads (still-valid #98 finding). Do not re-implement the postgres gate. * feat(store): rustls for production PostgreSQL sslmode=require (#100) * feat(store): rustls for production PostgreSQL sslmode=require Issue #80 remainder. sslmode=require/verify-ca/verify-full connect with rustls and Mozilla roots; certificates are always verified. allow/prefer are still rejected so the process cannot silently drop to plaintext. Live test against plaintext CI postgres proves fail-closed. Do not re-implement the postgres gate or the outbox. * docs: record PR #100 in the product-technical gap baseline * fix(store): rewrite verify-full sslmode for tokio-postgres 0.7 Still-valid #100 Devin finding. tokio-postgres 0.7 only parses disable/prefer/require. Map verify-ca/verify-full to require before connect; rustls still verifies certificates. Password query-lookalikes are left untouched. * feat(store): bound outbox listing and prune processed rows (#101) * feat(store): bound outbox listing and prune processed rows Still-valid #99 finding. GET /api/outbox returns at most EVENT_LIMIT rows (dead letters, then pending, then leased, then processed). Processed outbox_message rows prune to that cap; receipts and dead letters stay. Do not re-implement the outbox, postgres gate, or rustls. * docs: record PR #101 in the product-technical gap baseline * fix(store): prune processed outbox to EVENT_LIMIT on save and ack Still-valid #101 Devin finding. Snapshot save and worker ack used LIST_LIMIT (1000) while append used operator EVENT_LIMIT. Store the configured cap on PostgresPlane so all three paths retain the same processed-row bound. Receipts and dead letters stay. * feat(store): logical backup and isolated restore drill (#102) * feat(store): logical backup and isolated restore drill Issue #80 remainder stacked on #101. GET /api/backup exports a hashed tenant snapshot; POST /api/backup restores after schema and payload-hash checks; POST /api/backup/drill restores into an isolated tenant, compares unmasked invariants, and drops the drill rows. Declared RPO is last successful export; declared RTO is 60s. File/memory adapters report backup=disabled. Do not re-implement rustls, outbox, or retention. * docs: record PR #102 in the product-technical gap baseline * feat(store): non-owner PostgreSQL runtime role after migrate (#103) * feat(store): non-owner PostgreSQL runtime role after migrate Still-valid #98 finding. CI connects as a superuser, which bypasses FORCE RLS. Migrations stay on the login role, then SET ROLE wardnet_runtime (NOSUPERUSER, NOBYPASSRLS, not table owner). Missing tenant GUC yields no rows; DROP TABLE and DISABLE RLS are denied. Do not re-implement rustls, outbox, retention, or backup/restore. * docs: record PR #103 in the product-technical gap baseline * fix(store): restore logical backups across role-only schema versions v3 only provisions wardnet_runtime and does not change table shape. verify() accepts schema 2 through the current migration version so a role-only upgrade cannot void the last pre-upgrade logical backup. * feat(store): HASH-partition security_event by tenant (#104) * feat(store): HASH-partition security_event by tenant Convert unpartitioned security_event to PARTITION BY HASH (tenant_id) with eight children under pg_advisory_lock. Rows keep unmasked client IPs and paths. /healthz.event_partitions reports the child count. Logical restore still accepts schema 2 through the current version. * docs: record PR #104 in the product-technical gap baseline * feat(security): add bounded outbound fetch API (#113) * feat(security): add bounded outbound fetch API * fix(security): isolate fetch DNS pins per request * feat(security): route browser DNS and HTTPS through Wardnet (#116) * feat(security): add bounded outbound fetch API * feat(security): route browser DNS and HTTPS through Wardnet * fix(security): isolate fetch DNS pins per request * fix(dns): bound concurrent UDP query handling * fix(security): close destination policy review gaps * fix(runtime): make worker shutdown durable
seonghobae
added a commit
that referenced
this pull request
Aug 26, 2026
* feat(security): fail-closed destination policy for outbound HTTP One DestinationPolicy mediates gateway upstreams, threat-intel fetches, Clearfolio, SOC LLM, and the Coraza sidecar URL. Private, loopback, link-local, CGNAT, and metadata classes are denied unless DESTINATION_ALLOWLIST (or loopback development) permits them; DESTINATION_DENYLIST wins. Clients ignore ambient HTTP proxies and do not follow redirects. Refs #79. * docs: record PR #96 in the product-technical gap baseline * feat(security): harden destination policy per-IP CIDR and readiness order CIDR allowlist matches apply per resolved address, authorize non-default ports, and reject prefixes outside the address-family width. IPv6 site-local is a denied class. Hostnames that merely contain 0x are not hex IP literals. AppState constructors default to production policy; seeded fixtures opt into development. Blocking DNS runs on spawn_blocking with a timeout. Persistence and destination-list validation complete before the readiness line. * feat(security): pin outbound HTTP to evaluated destination addresses After destination policy allows a host, the reqwest client resolves only those IPs so a rebinding answer cannot reach loopback, private, or metadata classes. Host and SNI stay on the original name. Unpinned hostnames fail closed instead of falling back to OS DNS. * feat(waf): evaluate live gateway transactions with in-process libcoraza Issue #86 remainder: dlopen operator-supplied libcoraza and drive the C ABI on each /gateway request. Missing library or empty ruleset fail closed before bind. CI stays hermetic with a fixture cdylib that exports the same symbols. * docs: record PR #97 in the product-technical gap baseline * feat(store): require PostgreSQL as the production control plane Non-loopback binds fail closed without CONTROL_PLANE_DATABASE_URL. Migrations create 3NF two-word tables with default-deny RLS. Snapshot persist commits policy rows and audit records in one transaction. Loopback still uses the JSON file or memory adapter. * feat(store): transactional outbox and leased workers Issue #81 first slice on the PostgreSQL control plane. Security events append with an outbox row in one transaction instead of rewriting the snapshot. Workers claim with SKIP LOCKED, retry, dead-letter, and record unique receipts. Stdout SIEM is at-least-once; the receipt is the exactly-once ack. Also deterministic ORDER BY on postgres loads (still-valid #98 finding). Do not re-implement the postgres gate. * feat(store): rustls for production PostgreSQL sslmode=require (#100) * feat(store): rustls for production PostgreSQL sslmode=require Issue #80 remainder. sslmode=require/verify-ca/verify-full connect with rustls and Mozilla roots; certificates are always verified. allow/prefer are still rejected so the process cannot silently drop to plaintext. Live test against plaintext CI postgres proves fail-closed. Do not re-implement the postgres gate or the outbox. * docs: record PR #100 in the product-technical gap baseline * fix(store): rewrite verify-full sslmode for tokio-postgres 0.7 Still-valid #100 Devin finding. tokio-postgres 0.7 only parses disable/prefer/require. Map verify-ca/verify-full to require before connect; rustls still verifies certificates. Password query-lookalikes are left untouched. * feat(store): bound outbox listing and prune processed rows (#101) * feat(store): bound outbox listing and prune processed rows Still-valid #99 finding. GET /api/outbox returns at most EVENT_LIMIT rows (dead letters, then pending, then leased, then processed). Processed outbox_message rows prune to that cap; receipts and dead letters stay. Do not re-implement the outbox, postgres gate, or rustls. * docs: record PR #101 in the product-technical gap baseline * fix(store): prune processed outbox to EVENT_LIMIT on save and ack Still-valid #101 Devin finding. Snapshot save and worker ack used LIST_LIMIT (1000) while append used operator EVENT_LIMIT. Store the configured cap on PostgresPlane so all three paths retain the same processed-row bound. Receipts and dead letters stay. * feat(store): logical backup and isolated restore drill (#102) * feat(store): logical backup and isolated restore drill Issue #80 remainder stacked on #101. GET /api/backup exports a hashed tenant snapshot; POST /api/backup restores after schema and payload-hash checks; POST /api/backup/drill restores into an isolated tenant, compares unmasked invariants, and drops the drill rows. Declared RPO is last successful export; declared RTO is 60s. File/memory adapters report backup=disabled. Do not re-implement rustls, outbox, or retention. * docs: record PR #102 in the product-technical gap baseline * feat(store): non-owner PostgreSQL runtime role after migrate (#103) * feat(store): non-owner PostgreSQL runtime role after migrate Still-valid #98 finding. CI connects as a superuser, which bypasses FORCE RLS. Migrations stay on the login role, then SET ROLE wardnet_runtime (NOSUPERUSER, NOBYPASSRLS, not table owner). Missing tenant GUC yields no rows; DROP TABLE and DISABLE RLS are denied. Do not re-implement rustls, outbox, retention, or backup/restore. * docs: record PR #103 in the product-technical gap baseline * fix(store): restore logical backups across role-only schema versions v3 only provisions wardnet_runtime and does not change table shape. verify() accepts schema 2 through the current migration version so a role-only upgrade cannot void the last pre-upgrade logical backup. * feat(store): HASH-partition security_event by tenant (#104) * feat(store): HASH-partition security_event by tenant Convert unpartitioned security_event to PARTITION BY HASH (tenant_id) with eight children under pg_advisory_lock. Rows keep unmasked client IPs and paths. /healthz.event_partitions reports the child count. Logical restore still accepts schema 2 through the current version. * docs: record PR #104 in the product-technical gap baseline * feat(store): optimistic concurrency on postgres snapshots Issue #80 last remainder. tenant_account.snapshot_version must match the loaded token or persist fails closed (HTTP 409). Restores overwrite. File/memory stay single-writer. Do not re-implement rustls, outbox, runtime role, HASH, or backup/restore. * docs: record PR #105 in the product-technical gap baseline * fix(store): keep postgres snapshot_version aligned after startup save load_postgres was saving with OCC and leaving the in-memory token one behind the database, so every later management write returned HTTP 409. Advance the loaded snapshot_version to the value save() wrote. * feat(store): outbox consumers for TAXII, Clearfolio, and orchestrator Enqueue operator-triggered TAXII polls, Clearfolio submits, and contextual-orchestrator SOC analysis on the PostgreSQL leased outbox. Request path returns 202; GET /api/outbox/{id} exposes receipt evidence. Secrets stay in the credential registry. Startup postgres save advances snapshot_version so the first management write cannot false-conflict. * docs: record PR #106 in the product-technical gap baseline * feat(release): tagged GitHub Release with SHA-256 and immutable GHCR (#107) * feat(release): tagged GitHub Release with SHA-256 and immutable GHCR Issue #84 first slice. A vX.Y.Z tag builds a locked binary, checksums, a GitHub Release, and ghcr.io/contextualwisdomlab/waf-ids-ai-soc:vX.Y.Z with no moving latest tag. Promotion and rollback are tag-for-tag. Do not re-implement store slices or OCC. * docs: record PR #107 in the product-technical gap baseline * fix(release): basename checksums and serialize postgres GRANTs SHA256SUMS recorded dist/ prefixes so sha256sum -c failed next to the downloaded binary. Emit basenames. Parallel PostgresPlane connects raced HASH convert GRANT with SET ROLE GRANT (tuple concurrently updated); hold the advisory lock across both. Do not re-implement HASH layout. * feat(release): keyless cosign, SPDX SBOM, and SLSA on the same tag Issue #84 remainder. GitHub OIDC signs the binary, checksums, SBOMs, and the GHCR image by digest. Release is created only after signatures. Syft SPDX fails closed without syft or non-SPDX JSON. NIST SP 800-218 is attached. Do not re-implement checksums or store slices. * feat(release): refuse lightweight tags and pin k8s by digest Issue #84 remainder. Annotated vX.Y.Z tags only; lightweight tags fail closed before the release job builds. Kubernetes pin is the GHCR content digest; tag aliases are refused. Do not re-implement checksums or cosign/SBOM. * docs: record PR #109 in the product-technical gap baseline * feat(waf): detect OWASP CRS attack battery on the live binary Issue #11 first slice. The build-script libcoraza ABI stub gains a deterministic battery covering SQLi (942100), XSS (941100), path traversal (930100), Unix RCE (932100, with first-match ordering so '; cat /etc/passwd' attributes to RCE over traversal), and Log4j JNDI (944120) in raw and percent-encoded forms across URI and POST-body phases. tests/binary.rs now starts the real gateway with the stub engine, creates a block route through the admin API, fires nine cases over HTTP, and asserts each is 403-blocked citing the expected CRS rule id while a benign request still forwards; /api/events must record one event per attempt with the forwarded client IP kept unmasked. Doctoring: docs/doctoring/ci-attack-evidence-battery.md grounds the split between detection-path evidence (CI) and detection efficacy (operator-supplied libcoraza + Core Rule Set), APA 7th. * fix(control-plane): close OCC and credential race gaps * fix(release): capture pushed image digest
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Issue #80 remainder stacked on #101. Production PostgreSQL tenants get a hashed logical snapshot export, fail-closed restore, and an isolated restore drill.
GET /api/backup(admin read) exports policy, events, outbox, and receipts with a payload hash. Client IPs, paths, and actor names stay unmasked.POST /api/backup(admin write) restores after schema-version and payload-hash checks, then audits.POST /api/backup/drillrestores into an isolated tenant, compares semantic invariants, drops the drill tenant, and does not replace production./healthz.backupisreadyon PostgreSQL anddisabledon file/memory.on-demand-logical-snapshot). Declared RTO: 60 seconds.docs/papers/nist-sp-800-34r1-contingency-planning.pdf, public domain).Do not re-implement #78, sidecar #95, pin #96, libcoraza #97, postgres gate #98, outbox #99, rustls #100, or outbox retention #101.
Merge order: #94 independently; #95 then #96 then #97 then #98 then #99 then #100 then #101 then this. Do not
--adminmerge.Test plan
cargo fmt --checkcargo test --locked --workspacewith liveCONTROL_PLANE_TEST_DATABASE_URL(includespostgres_backup_restore_drill_preserves_unmasked_invariantsandbackup_verify_fails_closed_on_schema_and_hash)cargo clippy --locked --workspace --all-targets -- -D warnings/healthzplus/adminand/api/commercial/readinesssmokes