swarm-bao: let the controller write agent credentials, and gate that it can

Closes #4124.

The controller's policy granted only the bootstrap paths -- hive cert-auth
roles and hive ACLs. #4113 then made it a secret WRITER, and nothing related
the grants to the paths the code writes, so every matrix token provision
answered 403. The two halves landed on different issues and neither looked
wrong on its own.

`secret/data/` is KV v2's ACL prefix and is absent from the path the code
passes, so matching `swarm-secret-client`'s spelling literally would have
granted nothing. Write-only: the controller mints these and never reads one
back, and a read capability would let it recover every agent's credentials
rather than only replace them.

The gate is the point. Two module-eval arms -- the grant exists and is not
a broader wildcard, and its capability list is pinned whole, because an
ADDED capability is what a presence check misses -- plus a test in path.rs
pinning MOUNT/AGENT_PREFIX and naming the nix file, since renaming either
constant is a silent 403 rather than a compile error.

setup.md carried two warnings this makes false: that nothing in the tree had
ever authenticated to the store, and that no deployment shape mints a leaf
whose CN reads swarm-controller. glue-bao-tls.nix has minted one since #3726
item 1.
This commit is contained in:
atlas 2026-09-09 00:46:01 +02:00
commit cc8fb0ee44
4 changed files with 62 additions and 8 deletions

View file

@ -94,14 +94,12 @@ Put the token's value at `services.hyperhive.deploy.bao.bootstrapTokenFile`
store's container reads it, writes the `swarm-controller` policy, enables the
cert auth method, and creates the `swarm-controller` role that attaches the two.
⚠️ **None of this has been run against a live store.** Nothing in the tree has
ever authenticated to OpenBao, so treat the block above as derived rather than
exercised — the grants come from `-output-policy`, not from a swarm that came
up on them.
⚠️ **The role it creates has nothing to present a certificate for yet.** No
deployment shape mints a leaf whose CN reads `swarm-controller`, so until one
does, the role stays provisioning that waits for a consumer.
⚠️ **This has been exercised once, and it did not go all the way through.** The
first real provision against a live store returned 403 (hyperhive#4124), so
treat the block above as derived-then-partly-tested rather than proven: the
grants still come from `-output-policy`, not from a swarm that came up on them.
What that 403 does _not_ tell you is whether the login or the write was
refused — `bao read auth/cert/certs/swarm-controller` separates the two.
**Delete the file once that has run.** The unit skips when it's absent, so a
host that has finished bootstrapping stops carrying the credential — and the