8 Commits

Author SHA1 Message Date
monky-deployd-merger 35949a9246 Merge branch 'fix/reuse-lease-token' into 'main'
fix(agent): reuse a valid lease token; rate-limited lease no longer blocks a deploy; 0.1.6

See merge request tikali/applications/monky/monky-deployd!11
2026-09-07 20:27:03 +00:00
mdella a30014a096 fix(agent): reuse a valid lease token across applies; a rate-limited lease no longer blocks a deploy; 0.1.6
Every apply requested a new lease, so a failing deploy retried by the 60 s
timer burned tenancy's 5-leases-per-hour budget and then failed on
LEASE_RATE_LIMITED forever (env-qa-02 pilot). Now: reuse a lease-derived
token while lookup-self says it is valid; swap the bootstrap token for a
lease once; if tenancy rate-limits the lease while a working token exists,
apply with it and defer the swap.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KLB7jieMNRkTsJ2epr4Ds1
2026-09-07 20:25:38 +00:00
mdella 47c3840c95 Merge branch 'docs/attrs-now-added-by-broker' into 'main'
docs: the broker adds the agent attrs at register since 2026-09-07 — !9's 'not implemented yet' was stale on arrival

See merge request tikali/applications/monky/monky-deployd!10
2026-09-07 12:07:48 -07:00
Claude-Docs-Manager d0153dd71c docs: the broker DOES add the agent attrs at register since 2026-09-07 — !9's "not implemented yet" was stale on arrival
!9 (d34189c6) rewrote PROTOCOL.md §Where and how and README §Security model
to say monky-ziti does not add #monky-deploy-agent / #openbao-client yet and an
operator adds them by hand. That was verified against monky-ziti at the sweep
HEAD b44c50a4 (2026-09-05), not its live HEAD: monky-ziti ac258352 (2026-09-07
06:12Z, "feat(environments): host_extra_attrs — host-template-only attrs for
the agent's dial right", merged a82359dd) and monky-tenancy !29 9b84f9ee
(06:23Z, app/jobs_backends.py sends host_extra_attrs for non-k8s backends)
landed hours before !9 was opened. Both sentences now state what happens:
tenancy sends the attrs at register, the broker puts them on the host identity,
and only pre-!29 identities (env-qa-02's) were hand-added.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AW3QqEpwLV69KHn24Re45Q
2026-09-07 09:01:28 -07:00
mdella 6e8b2bf41d Merge branch 'docs/dd-0523-0525-0527-0528' into 'main'
docs+defaults: 443 everywhere the agent dials; broker attr not added yet; no jti state on the mount; release:gitea has not run (DD-0523/0525/0527/0528)

See merge request tikali/applications/monky/monky-deployd!9
2026-09-07 08:45:28 -07:00
Claude-Docs-Manager d34189c625 docs+defaults: 443 everywhere the agent dials; the attr the broker does not add yet; no jti state on the mount; release:gitea has not run
DD-0523 — !7 (31586c30, 0.1.5) moved install.sh and config.example.yaml to the
443 intercept after env-qa-02 hit "service not available". The ansible role
default (monky_deployd_tenancy_port) and TenancyCfg.port still said 8081, so
an ansible-installed box or a config that omits `port` still dialled the wrong
port; both now default to 443, the proxy-mapping and config tests follow, and
PROTOCOL.md §Where and how states the intercept port separately from the
in-pod 8081 and names openziti state/overlay/configs.json as the authority.

DD-0525 — PROTOCOL.md and README said "the broker adds the attr when the
identity is created at kit reveal". monky-ziti at b44c50a4 has no such code
(app/fabric.py host_identity_attrs carries the env template only) and openziti
docs/services.md says "Nothing carries the attr yet". Both now state the
dependency: an operator adds #monky-deploy-agent/#openbao-client on the
controller until the ADR-0028 addendum lands in monky-ziti.

DD-0527 — "the old kit's grant fails at login (unknown/used jti)". The
jwt-tenancy mount keeps no replay state (openbao terraform/jwt-tenancy.tf
see_env role: signature, aud, bound_claims, exp); a superseded grant logs in
until exp and the refusal is tenancy's 401 on the first bearer call. The
second-reveal paragraph, the grant-flow diagram and README §Security model say
so; FakeBao no longer pops a grant at login (the suite's superseded-token test
already goes through FakeTenancy.superseded_jtis, which is the real model).

DD-0528 — "Both locations keep being published": release:gitea has been a
never-run manual job on every tag pipeline (6999, 7044, 7066); README and
OPERATIONS.md now say when the Gitea mirror is published and that it has not
been yet.

Gates (local, py3.12): ruff format, ruff check, pytest 50 passed,
bash -n packaging/install.sh. `git grep 8081` afterwards hits only the in-pod
listener statements.

Doc-Drift: DD-0523 fixed
Doc-Drift: DD-0525 fixed
Doc-Drift: DD-0527 fixed
Doc-Drift: DD-0528 fixed
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AW3QqEpwLV69KHn24Re45Q
2026-09-07 08:22:02 -07:00
mdella b749a6d3b4 Merge branch 'docs/dd-0524-0526-protocol-shapes' into 'main'
docs(protocol): AppRole divergence is resolved history; lease example + fake follow tenancy's AgentLeaseOut (DD-0524/DD-0526)

See merge request tikali/applications/monky/monky-deployd!8
2026-09-06 23:23:32 -07:00
Claude-Docs-Manager 05724edde2 docs(protocol): the AppRole divergence is resolved history; the lease example and fake follow tenancy's AgentLeaseOut
DD-0524 — PROTOCOL.md §Divergences bullets 1-2 said monky-tenancy `main`
"still implements the AppRole lease and install kit" and that "the kit's
generated config uses tenancy.base_url". Both were false when the page was
published: tenancy !17 (288df791, merged 07:48Z) shipped AgentLeaseOut{env_id,
login_jwt, ttl_s, mount, role, addr}, the ES256 grant signer and the JWKS
sixteen minutes before v0.1.0 was tagged, and !22 (61bd0281) made the kit run
install.sh with flags instead of writing a config. The two bullets are now
dated "Resolved" notes; bullets 3-4 (report `detail`, X-Bundle-Sha) stand.

DD-0526 — the lease example sent `reason` and received a nested `vault{}`;
tenancy's AgentLease is `{env_id}` and AgentLeaseOut carries `addr` at the top
level (no vault object). The example now shows tenancy's shapes (with a note
that the agent still sends `reason` and tenancy ignores unknown fields), the
checkin example gains `auth_mount`/`auth_role` so it is the full AgentVaultOut,
tests/fakes.py emits `addr` the way tenancy does, and tenancy.py's docstring
and lease() read `addr` first (the `vault{}` fallback is kept so an older fake
or tenancy still leases). Verified against monky-tenancy app/schemas_backends.py
at 1fd51454 (AgentLease 296-297, AgentLeaseOut 300-309, AgentVaultOut 278-283).

Gates (local, py3.12): ruff format --check, ruff check, pytest 50 passed,
bash -n packaging/install.sh.

Doc-Drift: DD-0524 fixed
Doc-Drift: DD-0526 fixed
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AW3QqEpwLV69KHn24Re45Q
2026-09-06 23:13:09 -07:00
17 changed files with 136 additions and 51 deletions
+12
View File
@@ -1,8 +1,20 @@
<!-- xlate:verbatim-fences -->
# Changelog
## 0.1.6 — 2026-09-07
- **A valid lease token is reused across applies.** Every apply requested a new lease, so a failing deploy
retried by the 60 s timer burned tenancy's 5-leases-per-hour budget and then failed on
`LEASE_RATE_LIMITED` forever (env-qa-02 pilot). Now: reuse a lease-derived token while `lookup-self`
says it is valid; swap the bootstrap token for a lease once; and if tenancy rate-limits the lease while a
working token exists, apply with it and defer the swap instead of failing the deploy.
## 0.1.5 — 2026-09-07
- **`tenancy.port` defaults to 443 everywhere** — the ansible role default and the `TenancyCfg` default now
match `install.sh`/`config.example.yaml` (0.1.5 only changed those two); PROTOCOL.md states the intercept
port separately from the in-pod 8081, that the broker does not add `#monky-deploy-agent` yet, that the jwt
mount keeps no `jti` state, and that `release:gitea` has not run (doc-drift DD-0523/0525/0527/0528).
- **`${VAR}` inside comment lines is not a reference.** The renderer's `.env.template` header literally says
"substitutes every ${VAR}", which the refusal check counted as an unresolved variable
(`ENV_INCOMPLETE: unresolved: VAR`) — the first bundle on env-qa-02 was refused for it.
+10 -6
View File
@@ -23,9 +23,9 @@ you the enrolment JWT, a one-time **bootstrap deploy grant** and the read-only *
```sh
T=<deploy token> # read-only GitLab deploy token (read_package_registry); the kit carries it
curl -sSf -H "DEPLOY-TOKEN: $T" https://scm.tikali.ai/api/v4/projects/69/packages/generic/monky-deployd/0.1.5/install.sh \
curl -sSf -H "DEPLOY-TOKEN: $T" https://scm.tikali.ai/api/v4/projects/69/packages/generic/monky-deployd/0.1.6/install.sh \
| sudo bash -s -- --env env-qa-02 --site cbs --token "$T" --bootstrap-file bootstrap.jwt --enrol-jwt ./monky-host.env-qa-02.jwt
# [--transport sdk|proxy|system] [--version 0.1.5] [--docker-data-root /home/docker-data] [--laptop] [--bao-ca openbao-ca.pem] [--source gitlab|gitea]
# [--transport sdk|proxy|system] [--version 0.1.6] [--docker-data-root /home/docker-data] [--laptop] [--bao-ca openbao-ca.pem] [--source gitlab|gitea]
```
`install.sh` installs `ziti-edge-tunnel` (OpenZiti `jammy` suite) and `docker-compose-plugin` if
@@ -98,11 +98,13 @@ monky-deployd version
## Security model
- **Identity = the box's ziti host identity.** Only identities with `#monky-deploy-agent` can dial
tenancy's agent entrypoint; `#openbao-client` reaches OpenBao. The agent reads the identity
tenancy's agent entrypoint; `#openbao-client` reaches OpenBao — tenancy sends both to the broker
as `host_extra_attrs` at register (monky-tenancy !29 / monky-ziti `ac258352`, 2026-09-07), so a
backend registered since then dials without hand-added attrs (see PROTOCOL.md §Where and how). The agent reads the identity
through an ACL (`setfacl -m u:monky-deployd:r`), never owns it.
- **The bearer to tenancy is the agent's own OpenBao token**, minted by OpenBao from a
tenancy-signed ES256 deploy grant (`aud openbao-see-env`, `kind deploy-grant`, 1 h, single-use
`jti`). Tenancy verifies it with `auth/token/lookup`, pins `meta.env_id`, and refuses a token
tenancy-signed ES256 deploy grant (`aud openbao-see-env`, `kind deploy-grant`, 1 h; its `jti`
becomes the token's `meta.grant_jti` — the mount keeps no replay state, tenancy does). Tenancy verifies it with `auth/token/lookup`, pins `meta.env_id`, and refuses a token
whose `meta.grant_jti` was superseded (kit re-reveal, retire) → `401 AGENT_UNAUTHENTICATED`.
**No AppRole, nothing to unwrap** (Gate 1 result, 2026-09-05).
- **The agent never receives a secret from tenancy.** Bundles carry placeholders; the agent reads
@@ -160,7 +162,9 @@ blocking on `main`/tags (manual on MRs); the `.deb` always ships the SDK wheel.
`release` uploads to the GitLab generic package registry + release (**the primary download**; the
project is private, so the installer sends the read-only deploy token), and `release:gitea` publishes the same assets on the Gitea mirror (the `--source gitea`
alternative; automatic when `GITEA_TOKEN` is set, manual otherwise — see `docs/OPERATIONS.md` for
the by-hand recipe). Both locations keep being published.
the by-hand recipe). The GitLab registry is published by every tag pipeline; the Gitea mirror only
when `release:gitea` runs — automatically once `GITEA_TOKEN` is set in CI, by hand otherwise — so
check the Gitea release page before pointing an installer at it.
## See also
+1 -1
View File
@@ -16,7 +16,7 @@ template → optional openbao-ca PEM → optional `monky-deployd-proxy.service`
| var | note |
|---|---|
| `monky_deployd_version` | pinned release, e.g. `0.1.5` |
| `monky_deployd_version` | pinned release, e.g. `0.1.6` |
| `monky_deployd_download_token` | **vaulted**: GitLab deploy token, scope `read_package_registry` only (revocable) — the registry is private; seeded in OpenBao at `monky/monky-tenancy/deployd-download` key `token` (path/key are the operator's choice). Empty = no header (only works with the Gitea `base_url`) |
| `monky_deployd_env_id` / `_site` / `_transport` | per host (`env-dev-06`, `cbs`, `sdk`) |
| `monky_deployd_bootstrap_grant` | tenancy-minted deploy grant (1 h) — `ansible-vault` or a lookup at play time; empty keeps the existing token |
@@ -1,7 +1,7 @@
---
# monky_deployd — install and configure the Monky backend pull agent (MONKY-ADR-0028 §D).
# Copy this role into osg1-07 (roles/monky_deployd) and roll to env-dev-06..09 after the pilot.
monky_deployd_version: "0.1.5"
monky_deployd_version: "0.1.6"
monky_deployd_deb: "monky-deployd_{{ monky_deployd_version }}_amd64.deb"
# PRIMARY download = the GitLab project's generic package registry on scm.tikali.ai. Inside the estate
# gitea.cbs.tikali.net is split-horizon to jump1's RED EIP (10.10.0.175, no HTTP ingress), so backend
@@ -33,7 +33,7 @@ monky_deployd_disk_factor: 1.5
monky_deployd_disk_headroom_bytes: 2147483648
monky_deployd_tenancy_service: monky.tenancy.deploy
monky_deployd_tenancy_port: 8081
monky_deployd_tenancy_port: 443 # the intercept port of monky.tenancy.deploy (host.v1 forwards to 8081 in the pod)
monky_deployd_bao_addr: "https://bao.cbs.tikali.net:8200"
monky_deployd_bao_auth_mount: jwt-tenancy
monky_deployd_bao_role: see-env
+1 -1
View File
@@ -1,4 +1,4 @@
# /etc/monky-deployd/config.yaml — monky-deployd v0.1.5 (MONKY-ADR-0028 §D)
# /etc/monky-deployd/config.yaml — monky-deployd v0.1.6 (MONKY-ADR-0028 §D)
# Written by packaging/install.sh (or the ansible role monky_deployd). YAML *subset*: maps, scalars,
# simple lists, comments. Keys not listed here are a config error.
+2 -1
View File
@@ -108,7 +108,8 @@ download; xtrace is switched off). A 401 on the download means the token is miss
lacks the scope. Inside the estate `gitea.cbs.tikali.net` is split-horizon to jump1's RED EIP
(`10.10.0.175`), which has no HTTP ingress, so backend boxes cannot fetch from the Gitea mirror
(cbs/iac#102). Off-estate, `install.sh --source gitea` (ansible: `monky_deployd_base_url`/`_deb_url`,
no token) uses the Gitea release instead. The tag pipeline publishes to both (`release`, `release:gitea`).
no token) uses the Gitea release instead. The tag pipeline always publishes to GitLab (`release`); `release:gitea` is automatic only with
`GITEA_TOKEN` in CI and a manual job otherwise — it has not run on a tag pipeline yet.
## Publishing a release to Gitea by hand
+36 -19
View File
@@ -10,8 +10,15 @@ The agent side of the backend lifecycle agent protocol — [MONKY-ADR-0028](http
- **Where:** tenancy's **agent entrypoint** — a second container (`app.agent_main`,
`127.0.0.1:8081`) bound to the ziti service **`monky.tenancy.deploy`** by a `ziti-edge-tunnel
run-host` sidecar. Unreachable from the public ingress. Only host identities carrying
`#monky-deploy-agent` may dial (the broker adds the attr when the identity is created at kit reveal).
run-host` sidecar. Unreachable from the public ingress. **The agent dials the service's
INTERCEPT port, `tcp/443`** (`monky.tenancy.deploy.intercept.v1` in openziti
`state/overlay/configs.json` — the authority for the port); the sidecar's `host.v1` forwards to
`127.0.0.1:8081` inside the pod. `tenancy.port` in the config, `install.sh` and the ansible role all
default to 443 (0.1.5). Only host identities carrying `#monky-deploy-agent` may dial
(`monky.tenancy.deploy.dial`, openziti). Tenancy sends that attr and `#openbao-client` to the
broker as `host_extra_attrs` when it registers an agent-backed backend (monky-tenancy !29,
monky-ziti `ac258352` — both 2026-09-07), so `monky-host.<env>` carries them from creation;
identities created before that (env-qa-02's) had them added on the controller by hand.
- **Transport:** plain HTTP inside the mesh (`transport: sdk` dials by service name;
`proxy``127.0.0.1:18443`). OpenBao is the existing `openbao` ziti service (`#openbao-client`),
dialled as `https://bao.cbs.tikali.net:8200` (an intercept name), TLS validated against the
@@ -29,7 +36,7 @@ The agent side of the backend lifecycle agent protocol — [MONKY-ADR-0028](http
```
kit reveal (admin, once) tenancy signs a bootstrap DEPLOY GRANT: ES256 JWT
iss <tenancy issuer>, aud openbao-see-env, sub agent:<env_id>,
env_id, kind deploy-grant, jti (single-use, recorded), exp now+1h
env_id, kind deploy-grant, jti (→ token meta grant_jti; tenancy records it), exp now+1h
install.sh stages it at /etc/monky-deployd/bootstrap.jwt (0600)
first tick POST https://bao…/v1/auth/jwt-tenancy/login {"role":"see-env","jwt":<grant>}
→ auth.client_token (ttl 24 h, max 30 d, policy see-env,
@@ -49,10 +56,13 @@ OpenBao side (Terraform, `tikali/services/security/openbao`): mount `jwt-tenancy
`monky/data/{{identity.entity.aliases.<accessor>.metadata.env_id}}/see/*` read + metadata list +
`auth/token/{renew-self,lookup-self}`. **Gate 1 v2 passed 2026-09-05 07:40Z** (distinct entity per env).
**Second-reveal semantics.** A kit re-reveal (or a retire) supersedes every earlier grant of the
backend and revokes the token accessors tenancy knows. The old kit's grant fails at login
(unknown/used `jti`), and a token already minted from it is refused at the next call with
`401 AGENT_UNAUTHENTICATED` (its `meta.grant_jti` is superseded) — **not** `AGENT_ENV_MISMATCH`.
**Second-reveal semantics.** A kit re-reveal (or a retire) supersedes every earlier bootstrap grant
of the backend (tenancy marks their `backend_leases` rows revoked) and revokes the token accessors
tenancy knows. The `jwt-tenancy` mount keeps **no** replay state — it validates signature, `aud`,
`bound_claims` and `exp` only (openbao `terraform/jwt-tenancy.tf`, the `see_env` role) — so a
superseded grant still logs in until its 1 h `exp`; the token it yields, like a token already minted
from it, is refused by tenancy at the first bearer call with `401 AGENT_UNAUTHENTICATED` (its
`meta.grant_jti` is superseded) — **not** `AGENT_ENV_MISMATCH`.
The agent then deletes its token; if a fresh `bootstrap.jwt` is on disk it bootstraps again in the
same tick, otherwise it exits 1 and says "re-run the install kit".
@@ -70,7 +80,8 @@ same tick, otherwise it exits 1 and says "re-run the install kit".
```json
{"env_id": "env-qa-02", "desired_sha": "7a10…", "action": "apply", "purge_volumes": false,
"bundle_url": "/v1/agent/bundle/env-qa-02/7a10…", "checkin_interval_s": 60,
"vault": {"addr": "https://bao.cbs.tikali.net:8200", "mount": "monky", "prefix": "env-qa-02/see"}}
"vault": {"addr": "https://bao.cbs.tikali.net:8200", "mount": "monky", "prefix": "env-qa-02/see",
"auth_mount": "jwt-tenancy", "auth_role": "see-env"}}
```
`action`: `apply` (desired ≠ applied), `none` (converged → heartbeat), `down` (retire; `purge_volumes`
is only meaningful here). `vault.mount` is the **KV** mount; the agent adopts it if it differs from
@@ -109,12 +120,16 @@ address, pull, renderer, images_policy, secrets_provider, agent{…}, files[]`.
### `POST /v1/agent/lease`
```json
{"env_id": "env-qa-02", "reason": "apply"} // reason: apply | renew
{"env_id": "env-qa-02"}
```
(The agent also sends `"reason": "apply" | "renew"` for its own logs; tenancy's `AgentLease` is
`{env_id}` and ignores unknown fields.)
```json
{"env_id": "env-qa-02", "login_jwt": "eyJ…", "ttl_s": 3600, "mount": "jwt-tenancy", "role": "see-env",
"vault": {"addr": "https://bao.cbs.tikali.net:8200", "mount": "monky"}}
"addr": "https://bao.cbs.tikali.net:8200"}
```
`addr` is the OpenBao address for the login below, top-level (tenancy `AgentLeaseOut`); the KV
mount/prefix come from `checkin`'s `vault`, not from the lease.
Then `POST /v1/auth/{mount}/login {"role": "{role}", "jwt": "{login_jwt}"}` on OpenBao. A body with
`wrapping_token` / `role_id` (the pre-Gate-1 AppRole lease) is refused with `LEASE_SHAPE` → report
`failed`. `429 LEASE_RATE_LIMITED` → exit 75.
@@ -133,15 +148,17 @@ first 2 KiB land in the audit log — it has been through the redactor.
## Divergences (2026-09-05)
- **monky-tenancy `main` (MR !15) still implements the AppRole lease and install kit**
(`AgentLeaseOut{wrapping_token, role_id}`, `bootstrap.wrap`, `bao.approle` in the kit's config). The
binding design is the plan's Gate 1 RESULT / ADR-0028 amendment: `{login_jwt, ttl_s, mount, role}`
and `POST /v1/auth/jwt-tenancy/login`. This agent implements the latter; against an un-migrated
tenancy it reports `failed` with `LEASE_SHAPE` and refuses `bao.approle` in its config. The tenancy
follow-up (deploy-grant signer, JWKS, `lease` shape, kit → `bootstrap.jwt`) is tracked on
monky-tenancy.
- The kit's generated config uses `tenancy.base_url: http://monky.tenancy.deploy:8081` — accepted as
an alias for `tenancy.{scheme,host,port}`.
- **Resolved 2026-09-05 (before v0.1.0 was tagged):** monky-tenancy !17 (`288df791`) landed the
tenancy side of the Gate 1 RESULT / ADR-0028 amendment — `AgentLeaseOut{env_id, login_jwt, ttl_s,
mount, role, addr}`, the ES256 deploy-grant signer (`app/agent_keys.py`), `GET
/.well-known/agent-jwks.json`, no AppRole and no response wrapping anywhere in tenancy. The
`LEASE_SHAPE` refusal of a `wrapping_token`/`role_id` body stays in this agent as a guard against a
stale tenancy, not as a description of `main`.
- **Resolved 2026-09-05:** the kit no longer generates a config file. Since monky-tenancy !22
(`61bd0281`) the one-time install script stages the bootstrap grant and runs `install.sh --env …
--site … --version … --bootstrap-file …`, and `install.sh` writes `/etc/monky-deployd/config.yaml`
(tenancy service name, OpenBao address and `jwt-tenancy`/`see-env`). `tenancy.base_url:` remains an
accepted alias for `tenancy.{scheme,host,port}` (`config.py`) for hand-written configs.
- `report` gains an optional `detail` (doc 24 §3.3); tenancy's `AgentReport` ignores unknown fields
today — if `strict` bodies land, `detail` folds into `log_tail`.
- Bundle sha header: tenancy sends `X-Bundle-Sha`, doc 24 says `X-Bundle-Sha256`; the agent reads
+1 -1
View File
@@ -4,4 +4,4 @@ Dials monky-tenancy over the mesh with the box's host identity, fetches the rend
leases a deploy grant, logs in to OpenBao, reads its own secrets, runs `docker compose`,
reports. Stdlib only; the optional `openziti` SDK is the `sdk` transport."""
__version__ = "0.1.5"
__version__ = "0.1.6"
+24 -1
View File
@@ -277,7 +277,7 @@ class Agent:
# secrets: lease -> login -> reads (values never logged; names only)
entries = b.manifest.get("entries", [])
if entries:
token = self._lease_login("apply")
token = self._token_for_apply()
for e in entries:
self._values[e["var"]] = self.bao.kv_read(token, e["path"], cfg.env_id, e.get("version"))
log.info("read %d secret(s): %s", len(entries), ", ".join(sorted(self._values)))
@@ -392,6 +392,29 @@ class Agent:
if old and old != token:
self.bao.revoke_self(old)
def _token_for_apply(self) -> str:
"""The OpenBao token to read this bundle's secrets with. A lease-derived token that is
still valid is REUSED (a lease per apply burned tenancy's 5/h budget on every retry —
env-qa-02 pilot, 2026-09-07); the bootstrap token is swapped for a lease once; and if
tenancy rate-limits the lease while we hold a working token, apply with what we have
and swap later rather than fail the deploy."""
cfg = self.cfg
st = self.state.token
if self.token and st is not None and st.source == "lease":
try:
info = self.bao.lookup_self(self.token)
if int(info.get("ttl") or 0) > cfg.bao.renew_below_s:
return self.token
except BaoError as exc:
log.info("lease token no longer valid (%s); re-leasing", exc)
try:
return self._lease_login("apply")
except RateLimited as exc:
if self.token:
log.warning("lease rate-limited (%s); applying with the current token, swap deferred", exc)
return self.token
raise
def _lease_login(self, reason: str) -> str:
assert self.tenancy is not None
lease = self.tenancy.lease(reason)
+1 -1
View File
@@ -129,7 +129,7 @@ def parse_yaml_subset(text: str) -> dict:
class TenancyCfg:
service: str = "monky.tenancy.deploy"
host: str = "" # intercept host; defaults to `service`
port: int = 8081
port: int = 443 # the service's INTERCEPT port (openziti); the pod-side listener is 127.0.0.1:8081
scheme: str = "http"
proxy_addr: str = "127.0.0.1:18443"
timeout_s: int = 30
+4 -2
View File
@@ -4,7 +4,7 @@ Bearer = the agent's OpenBao token (from the `jwt-tenancy` login). Tenancy pins
the token's `meta.env_id` (403 AGENT_ENV_MISMATCH -> exit 78, never retried) and refuses a token
whose deploy grant was superseded (401 AGENT_UNAUTHENTICATED -> re-bootstrap or re-run the kit).
Lease shape of record (Gate 1 v2, 2026-09-05): `{login_jwt, ttl_s, mount, role, vault}`. An
Lease shape of record (Gate 1 v2, 2026-09-05): `{env_id, login_jwt, ttl_s, mount, role, addr}`. An
AppRole-era body (`wrapping_token`, `role_id`) is refused loudly there is nothing to unwrap."""
from __future__ import annotations
@@ -158,13 +158,15 @@ class TenancyClient:
"(ADR-0028 amendment 2026-09-05)",
)
raise TenancyError(200, "LEASE_SHAPE", "lease response carries no login_jwt")
# shape of record (tenancy AgentLeaseOut): `addr` is top-level; a pre-0.1.x `vault{}` object
# is still read as a fallback so an older fake or tenancy does not break the lease
v = js.get("vault") or {}
return Lease(
login_jwt=str(js["login_jwt"]),
ttl_s=int(js.get("ttl_s") or 3600),
mount=str(js.get("mount") or "jwt-tenancy"),
role=str(js.get("role") or "see-env"),
vault=Vault(addr=v.get("addr"), mount=v.get("mount"), prefix=v.get("prefix")),
vault=Vault(addr=js.get("addr") or v.get("addr"), mount=v.get("mount"), prefix=v.get("prefix")),
)
def report(
+3 -3
View File
@@ -3,9 +3,9 @@
#
# T=<read-only GitLab deploy token, scope read_package_registry> # from the install kit / OpenBao
# curl -sSf -H "DEPLOY-TOKEN: $T" \
# https://scm.tikali.ai/api/v4/projects/69/packages/generic/monky-deployd/0.1.5/install.sh \
# https://scm.tikali.ai/api/v4/projects/69/packages/generic/monky-deployd/0.1.6/install.sh \
# | sudo bash -s -- --env env-qa-02 --site cbs --token "$T" --bootstrap-file bootstrap.jwt \
# [--transport sdk|proxy|system] [--version 0.1.5] [--enrol-jwt /path/monky-host.env-qa-02.jwt] \
# [--transport sdk|proxy|system] [--version 0.1.6] [--enrol-jwt /path/monky-host.env-qa-02.jwt] \
# [--laptop] [--source gitlab|gitea] [--docker-data-root /home/docker-data]
#
# --token / MONKY_DEPLOYD_TOKEN: the GitLab project is PRIVATE (its parent groups are private, so it
@@ -29,7 +29,7 @@
set -euo pipefail
umask 077
DEFAULT_VERSION="0.1.5"
DEFAULT_VERSION="0.1.6"
# Download source. PRIMARY is the GitLab project's generic package registry on scm.tikali.ai: inside
# the estate gitea.cbs.tikali.net is split-horizon to jump1's RED EIP (10.10.0.175), which has no HTTP
# ingress, so backend boxes cannot reach the Gitea mirror (cbs/iac#102); scm.tikali.ai they can. The
+1 -1
View File
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
[project]
name = "monky-deployd"
version = "0.1.5"
version = "0.1.6"
description = "Monky backend pull agent: checkin -> bundle -> lease -> OpenBao -> docker compose -> report, over the ziti mesh (MONKY-ADR-0028)"
readme = "README.md"
requires-python = ">=3.12"
+6 -3
View File
@@ -190,9 +190,11 @@ class FakeBao:
return h._send(400, {"errors": ["error validating token: expired"]})
if body.get("role") != self.role:
return h._send(400, {"errors": [f"role {body.get('role')!r} could not be found"]})
meta = self.grants.pop(body.get("jwt", ""), None) # single-use grant
# the jwt mount keeps no jti/replay state: a known grant logs in until it expires;
# a SUPERSEDED grant is refused by tenancy (FakeTenancy.superseded_jtis), not here
meta = self.grants.get(body.get("jwt", ""))
if meta is None:
return h._send(400, {"errors": ["error validating token: unknown or already used grant"]})
return h._send(400, {"errors": ["error validating token: unknown grant"]})
m = json.loads(meta)
tok = self.mint(m["env_id"])
self.tokens[tok]["grant_jti"] = m["jti"]
@@ -354,7 +356,8 @@ class FakeTenancy:
"ttl_s": 3600,
"mount": self.bao.mount,
"role": self.bao.role,
"vault": {"addr": "https://bao.cbs.tikali.net:8200", "mount": self.kv_mount},
# tenancy AgentLeaseOut: `addr` is top-level; KV mount/prefix come from checkin
"addr": "https://bao.cbs.tikali.net:8200",
},
headers={"Cache-Control": "no-store"},
)
+28 -5
View File
@@ -107,7 +107,7 @@ def test_rollback_refused_unless_allowed(bootstrapped, tenancy, fake_docker):
tenancy.set_files(first)
assert tick(cfg) == EX_FAIL
assert tenancy.reports[-1]["result"] == "failed" and "ROLLBACK_REFUSED" in tenancy.reports[-1]["detail"]
assert len(tenancy.leases) == 2 # a refused bundle never leases
assert len(tenancy.leases) == 1 # a refused bundle never leases; the 2nd apply reused the lease token
tenancy.set_files(make_files(meta={"agent": {"allow_rollback": True}}))
assert tick(cfg) == EX_OK
@@ -246,10 +246,9 @@ def test_legacy_approle_lease_is_refused_loudly(bootstrapped, tenancy, fake_dock
assert "compose up" not in fake_docker.subcommands()
def test_lease_rate_limited_is_temporary(bootstrapped, tenancy, fake_docker):
cfg = bootstrapped
tenancy.lease_limit = 0
assert tick(cfg) == EX_TEMPFAIL
# (test_lease_rate_limited_is_temporary was retired in 0.1.6: a rate-limited lease is only a
# temporary failure when NO working token exists; with one, the agent applies and defers the swap —
# see test_rate_limited_lease_does_not_block_an_apply_when_a_token_exists.)
def test_unhealthy_after_up_reports_failed_with_compose_logs(bootstrapped, tenancy, fake_docker):
@@ -320,3 +319,27 @@ def test_write_private_mode(tmp_path):
p = tmp_path / "d" / "f"
statemod.write_private(p, b"x")
assert oct(p.stat().st_mode & 0o777) == "0o600" and not any(n.startswith(".f.") for n in os.listdir(p.parent))
def test_second_apply_reuses_the_lease_token(bootstrapped, tenancy, bao, fake_docker):
"""A valid lease-derived token is reused: a new bundle does NOT lease again (5/h budget —
env-qa-02 pilot: a retried deploy re-leased every 60 s and hit LEASE_RATE_LIMITED forever)."""
cfg = bootstrapped
assert tick(cfg) == EX_OK
assert len(tenancy.leases) == 1
tenancy.set_files(make_files(manifest=make_manifest(versions={"gemini_api_key": 1})))
assert tick(cfg) == EX_OK
assert len(tenancy.leases) == 1 # reused
st = statemod.load(cfg.state_path, ENV)
assert st.applied_sha == tenancy.desired_sha and st.token.source == "lease"
def test_rate_limited_lease_does_not_block_an_apply_when_a_token_exists(bootstrapped, tenancy, bao, fake_docker):
"""tenancy 429 on lease while the bootstrap token still works → apply with it, swap deferred."""
cfg = bootstrapped
tenancy.lease_limit = 0
assert tick(cfg) == EX_OK
assert tenancy.leases == []
st = statemod.load(cfg.state_path, ENV)
assert st.applied_sha == tenancy.desired_sha and st.token.source == "bootstrap"
assert tenancy.reports[-1]["result"] == "applied"
+1 -1
View File
@@ -93,7 +93,7 @@ def test_proxy_transport_refuses_unmapped_hosts():
cfg = from_dict({"env_id": "env-dev-06", "site": "cbs", "transport": "proxy"})
t = build(cfg)
assert t.mapping[("monky.tenancy.deploy", 8081)] == ("127.0.0.1", 18443)
assert t.mapping[("monky.tenancy.deploy", 443)] == ("127.0.0.1", 18443)
assert t.mapping[("bao.cbs.tikali.net", 8200)] == ("127.0.0.1", 18200)
try:
t.connect("example.com", 443, 1)
+3 -3
View File
@@ -10,7 +10,7 @@ transport: sdk
identity: /opt/openziti/etc/identities/monky-host.env-qa-02.json
tenancy:
service: monky.tenancy.deploy
base_url: http://monky.tenancy.deploy:8081
base_url: http://monky.tenancy.deploy:443
bao:
service: openbao
addr: https://bao.cbs.tikali.net:8200 # intercept, not public DNS
@@ -27,7 +27,7 @@ volumes_on_absent: keep
def test_yaml_subset_parses_nested_maps_and_types():
d = c.parse_yaml_subset(KIT)
assert d["env_id"] == "env-qa-02"
assert d["tenancy"]["base_url"] == "http://monky.tenancy.deploy:8081"
assert d["tenancy"]["base_url"] == "http://monky.tenancy.deploy:443"
assert d["bao"]["addr"] == "https://bao.cbs.tikali.net:8200"
assert d["interval_s"] == 60 and d["laptop_mode"] is False
@@ -46,7 +46,7 @@ def test_yaml_subset_refuses_flow_style_and_tabs():
def test_config_defaults_and_derivations():
cfg = c.from_dict(c.parse_yaml_subset(KIT))
assert cfg.tenancy.host == "monky.tenancy.deploy" and cfg.tenancy.port == 8081 and cfg.tenancy.scheme == "http"
assert cfg.tenancy.host == "monky.tenancy.deploy" and cfg.tenancy.port == 443 and cfg.tenancy.scheme == "http"
assert cfg.bao_url == ("https", "bao.cbs.tikali.net", 8200)
assert cfg.deploy_dir == "/var/lib/monky-deployd/env-qa-02"
assert cfg.compose_project == "monky-env-qa-02"