Add AUTH_STACK so the Auth tab follows the stack in force
Authentik is the likely destination and the page had Authelia and lldap wired in at every level, so the seam goes in now: the panels and every endpoint action route off one conf value, and a stack that cannot be driven yet says so rather than drawing controls with nothing behind.
This commit is contained in:
@@ -807,6 +807,7 @@ Saved into `master.conf`, which does not need to be opened by hand.
|
||||
|
||||
| Setting | Control | Where | What it does |
|
||||
|---|---|---|---|
|
||||
| `AI_OWNER_HOST` | a text box | in this section | Which node runs the model, the retrieval index and the bug store. Every other node borrows it over the mesh, so a box without a GPU still gets the assistant — it just does not get the AI tab. Defaults to host1 when unset or malformed: whoever builds the mesh is host1. |
|
||||
| `AI_ENABLED` | a switch | Settings tab | Fail-closed: anything other than the literal "true" means off. |
|
||||
| `AI_CONNECT_TIMEOUT` | a number box, in seconds | in this section | seconds — probe when resolving which node has Ollama |
|
||||
| `AI_REQUEST_TIMEOUT` | a number box, in seconds | in this section | seconds — must clear a cold model load |
|
||||
@@ -1139,6 +1140,19 @@ Saved into `master.conf`, which does not need to be opened by hand.
|
||||
|---|---|---|---|
|
||||
| `ARRAY_STOP_SCRIPTS` | a list, one entry per line | in this section | Scripts run by array_stopping.sh for a planned shutdown — stops everything cleanly in order. Run sequentially (foreground) — each must complete before the next starts. Order matters: user scripts first (prevents new ops), then data movement, then containers. |
|
||||
|
||||
## Auth Stack
|
||||
|
||||
Reachable from:
|
||||
|
||||
- Auth tab → Auth settings → *Auth Stack*
|
||||
- Settings tab → All settings → *Auth Stack*
|
||||
|
||||
Saved into `master.conf`, which does not need to be opened by hand.
|
||||
|
||||
| Setting | Control | Where | What it does |
|
||||
|---|---|---|---|
|
||||
| `AUTH_STACK` | a dropdown (authelia_lldap, authentik) | in this section | Mesh-wide rather than per host: auth is the one service the partnership treats as shared, owned by the mesh owner and consumed by everyone else, so two nodes disagreeing about which stack is in force would mean two different answers to "who is this person". "authelia_lldap" — Authelia for access rules, lldap for users and groups. Fully implemented. "authentik" — single stack for both. NOT IMPLE… |
|
||||
|
||||
## Backup Verify
|
||||
|
||||
Reachable from:
|
||||
@@ -1179,7 +1193,7 @@ Saved into `master.conf`, which does not need to be opened by hand.
|
||||
|
||||
| Setting | Control | Where | What it does |
|
||||
|---|---|---|---|
|
||||
| `BUG_REPORT_LOCAL_ENABLED` | a switch | in this section | LOCAL ON — reports go to your own Gitea (HOST1_BUG_REPORT_* in host1.conf) and stay there. They do NOT reach the Varaverk maintainer. LOCAL OFF — reports open a prefilled GitHub issue you submit under your own account. Nothing is ever transmitted automatically: the report is shown read-only and sending is a separate press. |
|
||||
| `BUG_REPORT_LOCAL_ENABLED` | a switch | in this section | LOCAL ON — reports go to your own Gitea (see HOSTN_BUG_REPORT_* in host*.conf) and stay there. They do NOT reach the Varaverk maintainer. Turn it on if you want your own backlog. LOCAL OFF — reports open a prefilled GitHub issue you submit under your own account. |
|
||||
| `BUG_REPORT_GITHUB_REPO` | a text box | in this section | — |
|
||||
|
||||
## Certificate Monitor
|
||||
@@ -1923,6 +1937,16 @@ Saved into `master.conf`, which does not need to be opened by hand.
|
||||
| `SYS_WATCHDOG_RUNAWAY_STRIKES` | a number box, in consecutive | in this section | consecutive cycles before warning |
|
||||
| `SYS_WATCHDOG_MDSTAT_ERROR_LIMIT` | a number box | in this section | new errors in one cycle before acting |
|
||||
|
||||
## Timed Mutes
|
||||
|
||||
Route: Settings tab → All settings → *Timed Mutes*
|
||||
|
||||
Saved into `master.conf`, which does not need to be opened by hand.
|
||||
|
||||
| Setting | Control | Where | What it does |
|
||||
|---|---|---|---|
|
||||
| `WATCHDOG_MUTE_MAX_HOURS` | a number box | in this section | A mute silences one container for a bounded time and then expires on its own. It exists because every other exemption here is permanent — SCAN_IGNORE is a conf edit, an intentional stop lasts until cleared — so a temporary problem gets a permanent workaround that nobody revisits. Capped so "temporary" is enforced rather than intended: a mute longer than this is refused. |
|
||||
|
||||
## Transcode Management
|
||||
|
||||
Reachable from:
|
||||
@@ -2090,5 +2114,5 @@ Do not describe a route — there is none, and that is the point.
|
||||
HOST1 and HOST2 are what detect_hosts() matches the local hostname against. Editing one here would change which machine this believes it is.
|
||||
- **SHARED HOST CONFIGURATION** (`master.conf`) — `DATA_DIR`, `DB_DIR`, `STATE_DIR`, `AI_DATA_DIR`, `CACHE_BACKUP_DIR`, `LOG_ARCHIVE_DIR`, `BACKUP_DIR`, `CONF_BACKUP_DIR`, `PERSISTENT_CONF_CACHE`, `ARR_CACHE_BACKUP_DIR`
|
||||
DATA_DIR is the single on-disk root — move it and the whole tree follows, including the state the running scripts are holding open.
|
||||
- **State Files** (`master.conf`) — `SYS_WATCHDOG_STATE_FILE`, `DOCKER_WATCHDOG_FAILED_FILE`, `DOCKER_WATCHDOG_INTENTIONAL_FILE`, `SYS_WATCHDOG_REBOOT_LOG`, `SYS_WATCHDOG_OOM_FILE`
|
||||
- **State Files** (`master.conf`) — `SYS_WATCHDOG_STATE_FILE`, `DOCKER_WATCHDOG_FAILED_FILE`, `DOCKER_WATCHDOG_INTENTIONAL_FILE`, `WATCHDOG_MUTE_FILE`, `SYS_WATCHDOG_REBOOT_LOG`, `SYS_WATCHDOG_OOM_FILE`
|
||||
Paths to live state databases, derived from STATE_DIR. Repointing one orphans the state a watchdog is mid-way through writing.
|
||||
|
||||
Reference in New Issue
Block a user