feat(ops): self-heal disk pressure instead of only alerting

The 5-minute disk probe now reclaims storage automatically: from 85% it runs the gentle age-windowed Docker prune, from 90% it drops the age windows (docker-prune.sh --force: all unused build cache and unreferenced images, all stopped containers) so a mount can never silently max out. Alerts still fire at 85/90/95% and their hint now points at non-Docker growth when reclaiming is not enough. Force mode is reserved for the worker; deploys keep the gentle mode. Volumes are off-limits in every path.

scripts/jobs-worker.ts

@@ -17,7 +17,11 @@ import {
 	healthDegraded,
 } from "../src/lib/services/alert";
 import { runCatalogExport } from "../src/lib/services/catalog-git-export";
-import { diskLevel, parseDfOutput } from "../src/lib/services/disk-usage";
+import {
+	DISK_THRESHOLDS,
+	diskLevel,
+	parseDfOutput,
+} from "../src/lib/services/disk-usage";
 import { invalidateNewsCache } from "../src/lib/services/news-cache";
 import { rcon } from "../src/lib/services/rcon";
 
@@ -133,6 +137,30 @@ async function checkDiskUsage(): Promise<void> {
 			});
 			await diskPressure(usage);
 		}
+
+		// Self-healing: never let a mount max out. Safe prune from the warning
+		// mark, forced prune (drop age windows) from the error mark. Cooldown is
+		// keyed separately so a stuck fill level does not re-prune every run
+		// while still escalating to the forced path on real pressure.
+		if (usage.percent >= DISK_THRESHOLDS.error) {
+			if (canAlert(`disk-prune-force-${usage.mount}`)) {
+				logger.warn("Disk near full — forcing Docker cache reclaim", {
+					module: "jobs",
+					mount: usage.mount,
+					percent: usage.percent,
+				});
+				await pruneDockerCache(true);
+			}
+		} else if (usage.percent >= DISK_THRESHOLDS.warning) {
+			if (canAlert(`disk-prune-${usage.mount}`)) {
+				logger.info("Disk at high water mark — running safe Docker prune", {
+					module: "jobs",
+					mount: usage.mount,
+					percent: usage.percent,
+				});
+				await pruneDockerCache(false);
+			}
+		}
 	}
 }
 
@@ -282,8 +310,10 @@ async function cleanupOldSessions(): Promise<void> {
 
 /** Host-side: reclaim Docker's unused cache (build cache, unreferenced images,
  * stopped containers). Volumes and in-use images are never touched. No-op when
- * docker or the prune script is unavailable. */
-async function pruneDockerCache(): Promise<void> {
+ * docker or the prune script is unavailable. With force=true the age windows
+ * are dropped (docker-prune.sh --force) so every unused byte is reclaimed —
+ * the emergency path for a nearly-full disk. */
+async function pruneDockerCache(force = false): Promise<void> {
 	const { access } = await import("node:fs/promises");
 	const { resolve } = await import("node:path");
 	const { spawn } = await import("node:child_process");
@@ -294,7 +324,9 @@ async function pruneDockerCache(): Promise<void> {
 		return;
 	}
 	await new Promise<void>((resolvePromise) => {
-		const child = spawn("bash", [script], { stdio: "ignore" });
+		const child = spawn("bash", [script, ...(force ? ["--force"] : [])], {
+			stdio: "ignore",
+		});
 		child.on("error", (err) =>
 			captureWorkerError(
 				err,