fix: validate live resource monitor ownership in status
Bind authenticated snapshots to controller PID/starttime and validate process identity and held inode anchors before reporting current check health. Preserve identical signed foreign replacements and fail closed when live ownership is unavailable. Co-authored-by: GPT-6 <gpt-6@openai.com>
This commit is contained in:
1 parent
4efc066625
commit
5c85f3f0b9
5 files changed
+252
-7
No files matched your search
@@ -62,6 +62,13 @@ unavailable ownership. Resource failures continue to other resources; ownership
|
||||
failure stops all probes. These checks have unit/loopback-fixture coverage only,
|
||||
not new remote acceptance or generic automated recovery/controller support.
|
||||
|
||||
Generic `managed resource-status` validates the authenticated snapshot against
|
||||
the live controller's PID/starttime, executable, command, UID, namespaces and
|
||||
held directory/key/state/singleton inodes. Cached check results alone cannot
|
||||
report current health: missing ownership or an identical signed replacement
|
||||
returns `BLOCKED` with `stale: true` without modifying the foreign file.
|
||||
`CHECKS_ONLY`/`HEALTHY` still do not mean candidate or production readiness.
|
||||
|
||||
## Production preparation (isolated clean-run acceptance passed)
|
||||
|
||||
The standalone deployment candidate is a **new isolated target**, never an
|
||||
|
||||
Reference in new issue
Block a user