Watch
0
0
Fork
You've already forked hyperhive
0

swarm-controller: read the queue client secret from the store, drop the file
Some checks were skipped
public bin cache / build + push to preem:grid (push) Has been skipped

The controller's OIDC client secret (client `swarm-controller`, used for
the queue connection, the auth-bridge bearer and the OTLP push) came from
an operator-placed file, `deploy.swarm-controller.queue.clientSecretFile`,
handed in by `LoadCredential=`.

Now `swarm-secret-publish`, which already copies authelia's minted OIDC
secrets into the store, also publishes this one, to
`swarm/controller/swarm-controller/oidc/client`. That path sits under
`controller/`, which no hive's policy reads. The controller reads it once
at start with its existing store certificate and holds it in memory, as
`swarm_queue_client::ClientSecret::Value`. If the store is down, it
retries for about a minute and then fails the start, so `Restart=` tries
again.

Policy delta: the controller gets `read` on that leaf, and the publisher
gets `create`/`update` on that leaf.

Removed: the `queue.clientSecretFile` option (both spellings, now removed
options with a message), its singleHostSwarm default, the credential and
placeholder, and the path watcher plus its restart oneshot. A controller
without a store identity is now an eval error, because it has no other
way to get the secret.
This commit is contained in:
atlas 2026-09-28 19:00:31 +02:00
commit e94406cdb9
22 changed files with 595 additions and 247 deletions

View file

@ -364,18 +364,22 @@ let
!(baoGrantHere.containers.swarm-bao.config.systemd.services ? swarm-bao-secret-publisher-policy);
}
{
# The whole point of a second principal. The two prefixes it publishes to
# and not `swarm/`, so it cannot touch an agent's credentials; and no
# `read`, so a unit whose job is copying a file cannot recover what is
# already there. Pinned as the full capability list per prefix, because an
# added capability is exactly what a presence check misses.
name = "the publisher's grant is write-only and reaches the hive and service prefixes alone";
# The whole point of a second principal. The two prefixes and one leaf it
# publishes to and not `swarm/`, so it cannot touch an agent's
# credentials or the appservice token beside the controller's client;
# and no `read`, so a unit whose job is copying a file cannot recover
# what is already there. Pinned as the full capability list per path,
# because an added capability is exactly what a presence check misses.
name = "the publisher's grant is write-only and reaches the hive and service prefixes and the controller's client alone";
ok =
let
s = baoGrantHere.systemd.services.swarm-bao-secret-publisher-policy.script;
in
lib.hasInfix "path \"secret/data/swarm/hives/*\" {\n capabilities = [\"create\", \"update\"]" s
&& lib.hasInfix "path \"secret/data/swarm/services/*\" {\n capabilities = [\"create\", \"update\"]" s
&& lib.hasInfix "path \"secret/data/swarm/controller/swarm-controller/oidc/client\" {\n capabilities = [\"create\", \"update\"]\n}" s
&& !(lib.hasInfix "secret/data/swarm/controller/*" s)
&& !(lib.hasInfix "appservice-token" s)
&& !(lib.hasInfix "secret/data/swarm/agents" s)
&& !(lib.hasInfix "secret/data/swarm/*" s)
&& !(lib.hasInfix "sys/policies/acl" s);
@ -1114,6 +1118,13 @@ let
name = "the controller reads the swarm appservice token and cannot write it";
ok = lib.hasInfix "path \"secret/data/swarm/controller/swarm-controller/matrix/appservice-token\" {\n capabilities = [\"read\"]\n}" baoGrantHere.systemd.services.swarm-bao-controller-policy.script;
}
{
# The controller's own OIDC client secret, which the publisher writes and
# the controller reads at start. Pinned as the whole stanza, so an added
# capability fails.
name = "the controller reads its own client secret and cannot write it";
ok = lib.hasInfix "path \"secret/data/swarm/controller/swarm-controller/oidc/client\" {\n capabilities = [\"read\"]\n}" baoGrantHere.systemd.services.swarm-bao-controller-policy.script;
}
{
# Every role lives under a mount nothing else creates, and the granter
# holds no `sys/auth`, so the token-holding unit creates it — otherwise