fleetd #500: stop a wrong-interpreter or failed-parse reading a policy as empty

probe-member-credentials.sh used mapfile < <(producer) to parse the fetched policy. That
hides a producer failure three ways: mapfile is bash 4+ and missing on macOS's /bin/bash
3.2, a process substitution's exit status is never propagated to mapfile, and the
downstream reads (":-" defaults and a slice) never fire set -u on a short or unset array.
All three converge on the same "0 known names" refusal, which blames the policy for a
failure that is actually the interpreter or the parser.

Three distinct guards, each closing one cause with its own message:
- a BASH_VERSINFO gate at the top refuses outright on bash < 4 (exit 3)
- the parser's output is captured via command substitution instead of mapfile < <(...),
  so a non-zero jq/python3 exit is caught at the call while the fact still exists (exit 4)
- an arity check before the field slice refuses a parse that exits 0 but returns fewer
  than 5 fields (exit 5)

The existing "0 known names" guard is now honest: by the time it fires, the three causes
above are already ruled out, so it really does mean the policy has 0 known names.

fleetd/src/main/java/dev/ltms/fleet/auth/CallerResolver.java

@@ -244,7 +244,15 @@ public final class CallerResolver {
         // already names what happens if that case is handed the primary role: a worker→primary
         // escalation. So an unresolved caller is refused (ANONYMOUS — the same clean, already-tested
         // "authenticated as nothing" outcome used everywhere else in this method), never promoted.
-        return isLoopback(remoteAddr) && c.resolved() ? Principal.primary(c.pid()) : Principal.anonymous();
+        //
+        // fleetd #505: the OTHER way a real pid can wrongly reach here with a null terminal — not a
+        // failed lsof lookup, but a herdr error partway through PaneLocator's pane scan. c.resolved()
+        // says nothing about that; it only tests the lsof sentinel (by design — see
+        // ConnectionIdentity.Caller#resolved). c.scanComplete() is the separate signal: a scan that
+        // could not check every pane must not be read as "checked everywhere, no match" — the pane it
+        // could not check might have been the caller's own. So both must hold before this promotes.
+        return isLoopback(remoteAddr) && c.resolved() && c.scanComplete()
+                ? Principal.primary(c.pid()) : Principal.anonymous();
     }
 
     private boolean presentedTokenMatches(String authorizationHeader) {

fleetd/src/main/java/dev/ltms/fleet/herdr/PaneLocator.java

@@ -1,6 +1,8 @@
 package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.util.LinkedHashSet;
 import java.util.List;
@@ -36,6 +38,8 @@ import java.util.Set;
  */
 public final class PaneLocator {
 
+    private static final Logger log = LoggerFactory.getLogger(PaneLocator.class);
+
     /**
      * Bound on how many ancestor generations {@link #ancestorsOf} walks. This runs on every MCP
      * call, so a cycle or a pathologically deep process tree must not hang identity resolution;
@@ -73,22 +77,46 @@ public final class PaneLocator {
     }
 
     /**
-     * The {@code terminal_id} of the agent pane whose process tree contains {@code pid}, or
-     * {@code null} if no agent pane on any searched daemon owns it (e.g. the caller is the
-     * primary, or off-host).
+     * The outcome of a {@link #terminalForPid} scan: the {@code terminal_id} of the agent pane
+     * whose process tree contains the pid ({@link #terminal} is {@code null} if none matched),
+     * and whether the scan that produced that answer ran to completion on every daemon searched.
+     *
+     * <p>{@link #complete} is {@code false} exactly when some {@code pane.process_info} call
+     * failed and, despite that, no pane was ever found to own the pid. In that case a {@code null}
+     * {@link #terminal} means "could not tell", not "definitely not a worker" — fleetd #505: a
+     * transient herdr error on the very pane that <em>does</em> own the caller's pid must not read
+     * as a clean negative and fall through to {@code Principal.primary}, the same way #317's
+     * {@code Caller.resolved()} already guards a failed lsof lookup. Callers ({@code
+     * ConnectionIdentity}, {@code CallerResolver}) must refuse rather than promote on an incomplete
+     * scan.
+     *
+     * <p>When a pane genuinely owns the pid, {@link #complete} is {@code true} regardless of
+     * whether some other, unrelated pane failed to answer earlier in the same scan — a positive
+     * match is definitive and does not need every pane to have been checked (a pane that "vanished
+     * mid-scan" but was never the match is still a clean, complete result).
      */
-    public String terminalForPid(long pid) {
+    public record Lookup(String terminal, boolean complete) {
+        private static final Lookup NOT_FOUND = new Lookup(null, true);
+    }
+
+    /**
+     * Resolve {@code pid} to the agent pane whose process tree contains it, across every searched
+     * herdr daemon. See {@link Lookup} for how to read a {@code null} terminal.
+     */
+    public Lookup terminalForPid(long pid) {
         if (pid <= 0) {
-            return null;
+            return Lookup.NOT_FOUND;
         }
         Set<Long> ancestry = ancestorsOf(pid);
-        for (HerdrClient herdr : herdrs) {
-            String terminal = terminalForPid(herdr, ancestry);
-            if (terminal != null) {
-                return terminal;
+        boolean complete = true;
+        for (int i = 0; i < herdrs.size(); i++) {
+            Lookup outcome = scan(herdrs.get(i), i, herdrs.size(), ancestry);
+            if (outcome.terminal() != null) {
+                return outcome; // a definite match — no need to finish checking other clients
             }
+            complete = complete && outcome.complete();
         }
-        return null;
+        return new Lookup(null, complete);
     }
 
     /**
@@ -117,31 +145,51 @@ public final class PaneLocator {
         return ancestry;
     }
 
-    private static String terminalForPid(HerdrClient herdr, Set<Long> ancestry) {
+    /** Whether a pane owns one of the scanned pid's ancestors, or the check of it failed outright. */
+    private enum Ownership { OWNS, DOES_NOT_OWN, UNKNOWN }
+
+    private static Lookup scan(HerdrClient herdr, int clientIndex, int clientCount, Set<Long> ancestry) {
+        boolean complete = true;
         for (JsonNode pane : herdr.call("pane.list", Map.of()).path("panes")) {
             String paneId = pane.path("pane_id").asText(null);
-            if (paneId != null && paneOwnsAnyOf(herdr, paneId, ancestry)) {
-                return pane.path("terminal_id").asText(null);
+            if (paneId == null) {
+                continue;
+            }
+            Ownership owns = paneOwnsAnyOf(herdr, clientIndex, clientCount, paneId, ancestry);
+            if (owns == Ownership.OWNS) {
+                return new Lookup(pane.path("terminal_id").asText(null), true);
+            }
+            if (owns == Ownership.UNKNOWN) {
+                complete = false;
             }
         }
-        return null;
+        return new Lookup(null, complete);
     }
 
-    private static boolean paneOwnsAnyOf(HerdrClient herdr, String paneId, Set<Long> ancestry) {
+    private static Ownership paneOwnsAnyOf(HerdrClient herdr, int clientIndex, int clientCount,
+                                            String paneId, Set<Long> ancestry) {
         JsonNode info;
         try {
             info = herdr.call("pane.process_info", Map.of("pane_id", paneId)).path("process_info");
         } catch (HerdrException e) {
-            return false; // pane vanished mid-scan — just skip it
+            // fleetd #505: this used to be read as a clean "does not own it" (the pane vanished
+            // mid-scan, just skip it) — one boolean carrying two different facts. It is UNKNOWN
+            // now: if THIS pane is the one that owns the pid, the caller must not be told "no pane
+            // owns it", because that reads as a real primary and is promoted under loopback-trust.
+            log.warn("pane.process_info failed for pane {} on herdr client {} of {} during a "
+                            + "pid-owner scan — treating it as \"could not tell\", not a clean "
+                            + "negative (fleetd #505): {}",
+                    paneId, clientIndex + 1, clientCount, e.getMessage());
+            return Ownership.UNKNOWN;
         }
         if (ancestry.contains(info.path("shell_pid").asLong(-1))) {
-            return true;
+            return Ownership.OWNS;
         }
         for (JsonNode p : info.path("foreground_processes")) {
             if (ancestry.contains(p.path("pid").asLong(-1))) {
-                return true;
+                return Ownership.OWNS;
             }
         }
-        return false;
+        return Ownership.DOES_NOT_OWN;
     }
 }

fleetd/src/main/java/dev/ltms/fleet/mcp/ConnectionIdentity.java

@@ -33,9 +33,10 @@ public final class ConnectionIdentity {
 
     /**
      * The caller resolved from the connection: its worker {@code terminal} (or {@code null} for the
-     * primary / an off-host client) and its {@code pid} (or {@code -1} if not resolvable).
+     * primary / an off-host client), its {@code pid} (or {@code -1} if not resolvable), and whether
+     * the pane scan behind {@code terminal} ran to completion ({@link #scanComplete}).
      */
-    public record Caller(String terminal, long pid) {
+    public record Caller(String terminal, long pid, boolean scanComplete) {
 
         /**
          * Whether the OS peer-PID lookup actually succeeded — {@code false} means {@code pid} is
@@ -51,6 +52,10 @@ public final class ConnectionIdentity {
          * {@link ConnectionIdentity#isLoopback} is centralised rather than left for each caller to
          * reimplement: a raw {@code pid > 0} check duplicated at every call site is precisely the
          * "one rule, two copies" shape that let #305 drift.
+         *
+         * <p>This method is deliberately NOT widened for fleetd #505's failure (a herdr error
+         * during the pane scan, not a failed lsof lookup) — it still tests only the sentinel it is
+         * named for. #505 is a different axis, carried separately in {@link #scanComplete}.
          */
         public boolean resolved() {
             return pid > 0;
@@ -60,10 +65,11 @@ public final class ConnectionIdentity {
     /** Resolve the caller's terminal and PID from one peer-PID lookup. */
     public Caller resolve(String remoteAddr, int remotePort) {
         if (!isLoopback(remoteAddr)) {
-            return new Caller(null, -1); // only same-host callers can be workers
+            return new Caller(null, -1, true); // only same-host callers can be workers
         }
         long pid = pids.pidForLocalPort(remotePort);
-        return new Caller(panes.terminalForPid(pid), pid);
+        PaneLocator.Lookup lookup = panes.terminalForPid(pid);
+        return new Caller(lookup.terminal(), pid, lookup.complete());
     }
 
     /**

fleetd/src/test/java/dev/ltms/fleet/auth/CallerResolverTest.java

@@ -186,6 +186,43 @@ class CallerResolverTest {
         assertEquals(Role.PRIMARY, r.resolve("127.0.0.1", 99, "BEARER s3cret").role());
     }
 
+    // ── fleetd #505: a herdr error DURING THE SCAN must not be conflated with "not a worker" ──────
+    // #317 (above) covers a failed lsof lookup. This is the other input to the same decision: the
+    // lsof lookup succeeds (a real pid), but PaneLocator's own pane scan hits a herdr error on the
+    // pane that owns that pid — so c.resolved() is true and c.terminal() is null, exactly like a
+    // real primary. c.scanComplete() is what tells them apart.
+
+    /**
+     * The discriminating case named in the ticket: the error must land on the pane that DOES own
+     * the caller's pid, or the test proves nothing (any other pane's failure is invisible to the
+     * scan's outcome, since a match found elsewhere is definitive regardless).
+     */
+    @Test
+    void aHerdrErrorOnTheOwningPaneDuringTheScanIsRefusedNotPromotedToPrimary() {
+        FakeHerdr failing = new FakeHerdr().processInfoFailsForPane("w2:p7", "transient");
+        ConnectionIdentity incomplete = new ConnectionIdentity(new PaneLocator(failing), _ -> FakeHerdr.WORKER_PID);
+
+        Principal p = new CallerResolver(incomplete).resolve("127.0.0.1", 55555, null);
+
+        assertEquals(Role.ANONYMOUS, p.role(),
+                "an incomplete pane scan must never be read as a clean negative and promoted to primary");
+    }
+
+    /**
+     * The companion invariant: a herdr error on a DIFFERENT, non-owning pane must not turn every
+     * mid-scan teardown into a refusal — the real match is still found and resolves as a worker.
+     */
+    @Test
+    void aHerdrErrorOnANonOwningPaneStillResolvesTheRealWorker() {
+        FakeHerdr vanishedElsewhere = new FakeHerdr().processInfoFailsForPane("w2:p9", "pane_not_found");
+        ConnectionIdentity id = new ConnectionIdentity(new PaneLocator(vanishedElsewhere), _ -> FakeHerdr.WORKER_PID);
+
+        Principal p = new CallerResolver(id).resolve("127.0.0.1", 55555, null);
+
+        assertEquals(Role.WORKER, p.role());
+        assertEquals("term_a", p.terminal());
+    }
+
     @Test
     void aNonLoopbackCallerIsNeverThePrimaryUnderLoopbackTrust() {
         // Defence in depth: startup already refuses this pairing (validateAuthExposure), but if a

fleetd/src/test/java/dev/ltms/fleet/herdr/FakeHerdr.java

@@ -44,6 +44,7 @@ public final class FakeHerdr implements HerdrClient {
     private int workerTabPaneCount = 1;
     private String paneCloseErrorCode = null;
     private final Map<String, String> paneCloseErrorCodeFor = new ConcurrentHashMap<>();
+    private final Map<String, String> processInfoErrorCodeFor = new ConcurrentHashMap<>();
     private String tabCloseErrorCode = null;
     private final Map<String, String> tabCloseErrorCodeFor = new ConcurrentHashMap<>();
     private String agentSendErrorCode = null;
@@ -144,6 +145,18 @@ public final class FakeHerdr implements HerdrClient {
         return this;
     }
 
+    /**
+     * Make {@code pane.process_info} fail with this herdr error code, but only for the given
+     * {@code pane_id} — every other pane's {@code pane.process_info} still succeeds. Models a
+     * transient herdr failure partway through a {@link PaneLocator} pid→pane scan (fleetd #505):
+     * the scan must be able to tell "this pane does not own the pid" apart from "the scan could
+     * not check this pane at all", instead of collapsing both into one {@code false}.
+     */
+    public FakeHerdr processInfoFailsForPane(String paneId, String code) {
+        this.processInfoErrorCodeFor.put(paneId, code);
+        return this;
+    }
+
     /** Set the {@code agent_status} that {@code agent.get} reports (drives the injector). */
     public FakeHerdr agentStatus(String status) {
         this.agentStatus = status;
@@ -407,6 +420,13 @@ public final class FakeHerdr implements HerdrClient {
                           {"pane_id":"w2:p9","terminal_id":"term_shell","workspace_id":"w2","tab_id":"w2:t8"}]}""");
                 case "pane.process_info" -> {
                     Object paneId = params instanceof java.util.Map<?, ?> m ? m.get("pane_id") : null;
+                    String failCode = paneId == null ? null
+                            : processInfoErrorCodeFor.get(String.valueOf(paneId));
+                    if (failCode != null) {
+                        throw new HerdrException(
+                                "herdr error [" + failCode + "]: pane.process_info failed",
+                                failCode, null);
+                    }
                     yield "w2:p7".equals(paneId)
                             ? mapper.readTree(("""
                                 {"type":"pane_process_info","process_info":{"pane_id":"w2:p7","shell_pid":%d,

fleetd/src/test/java/dev/ltms/fleet/herdr/PaneLocatorContractTest.java

@@ -37,7 +37,7 @@ class PaneLocatorContractTest {
                         .path("pane").path("terminal_id").asText(null);
                 assertNotNull(terminalId, "seed pane should carry a terminal_id");
 
-                assertEquals(terminalId, new PaneLocator(herdr).terminalForPid(shellPid),
+                assertEquals(terminalId, new PaneLocator(herdr).terminalForPid(shellPid).terminal(),
                         "a real PID must resolve back to its own pane's terminal_id");
             } finally {
                 spaces.closeTab(tab.tab().tabId());

fleetd/src/test/java/dev/ltms/fleet/herdr/PaneLocatorTest.java

@@ -15,18 +15,20 @@ class PaneLocatorTest {
 
     @Test
     void resolvesTerminalForAForegroundPid() {
-        assertEquals("term_a", loc.terminalForPid(FakeHerdr.WORKER_PID));
+        assertEquals("term_a", loc.terminalForPid(FakeHerdr.WORKER_PID).terminal());
     }
 
     @Test
     void nullForAPidInNoPane() {
-        assertNull(loc.terminalForPid(999_999));
+        PaneLocator.Lookup outcome = loc.terminalForPid(999_999);
+        assertNull(outcome.terminal());
+        assertTrue(outcome.complete(), "a full, error-free scan that finds no match is complete");
     }
 
     @Test
     void nullForNonPositivePid() {
-        assertNull(loc.terminalForPid(0));
-        assertNull(loc.terminalForPid(-1));
+        assertNull(loc.terminalForPid(0).terminal());
+        assertNull(loc.terminalForPid(-1).terminal());
     }
 
     // --- two-daemon fallback (CB-185) -----------------------------------------
@@ -38,7 +40,7 @@ class PaneLocatorTest {
         HerdrClient lead = new FakeHerdr().withNoPanes();
         HerdrClient member = new FakeHerdr();
         PaneLocator two = new PaneLocator(lead, member);
-        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID));
+        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID).terminal());
     }
 
     @Test
@@ -48,13 +50,13 @@ class PaneLocatorTest {
         HerdrClient lead = new FakeHerdr();
         HerdrClient member = new FakeHerdr().withNoPanes();
         PaneLocator two = new PaneLocator(lead, member);
-        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID));
+        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID).terminal());
     }
 
     @Test
     void nullWhenNeitherClientHasTheMatch() {
         PaneLocator two = new PaneLocator(new FakeHerdr().withNoPanes(), new FakeHerdr().withNoPanes());
-        assertNull(two.terminalForPid(FakeHerdr.WORKER_PID));
+        assertNull(two.terminalForPid(FakeHerdr.WORKER_PID).terminal());
     }
 
     @Test
@@ -63,7 +65,7 @@ class PaneLocatorTest {
         // must behave exactly like the one-arg constructor, including making only one herdr call.
         FakeHerdr shared = new FakeHerdr();
         PaneLocator two = new PaneLocator(shared, shared);
-        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID));
+        assertEquals("term_a", two.terminalForPid(FakeHerdr.WORKER_PID).terminal());
         long paneListCalls = shared.calls.stream().filter(c -> c.method().equals("pane.list")).count();
         assertEquals(1, paneListCalls, "same-object lead/member must scan exactly once, not twice");
     }
@@ -75,7 +77,7 @@ class PaneLocatorTest {
         // Regression: a pid with no parent chain at all — no ancestry walk is needed to match it.
         OnePaneHerdr pane = new OnePaneHerdr("term_x", "pX", 5000, 6000);
         PaneLocator loc = new PaneLocator(pane, new FakeParentResolver());
-        assertEquals("term_x", loc.terminalForPid(5000));
+        assertEquals("term_x", loc.terminalForPid(5000).terminal());
     }
 
     @Test
@@ -83,7 +85,7 @@ class PaneLocatorTest {
         // Regression: same as above, but matching via the foreground-processes list.
         OnePaneHerdr pane = new OnePaneHerdr("term_x", "pX", 5000, 6000);
         PaneLocator loc = new PaneLocator(pane, new FakeParentResolver());
-        assertEquals("term_x", loc.terminalForPid(6000));
+        assertEquals("term_x", loc.terminalForPid(6000).terminal());
     }
 
     @Test
@@ -97,7 +99,7 @@ class PaneLocatorTest {
                 .parent(7002, 7001)   // grandchild -> child
                 .parent(7001, 5000);  // child -> shell (the pane's shell_pid)
         PaneLocator loc = new PaneLocator(pane, parents);
-        assertEquals("term_x", loc.terminalForPid(7002));
+        assertEquals("term_x", loc.terminalForPid(7002).terminal());
     }
 
     @Test
@@ -110,7 +112,7 @@ class PaneLocatorTest {
                 .parent(9002, 9001)
                 .parent(9001, 9000); // chain never reaches 5000 or 6000
         PaneLocator loc = new PaneLocator(pane, parents);
-        assertNull(loc.terminalForPid(9002));
+        assertNull(loc.terminalForPid(9002).terminal());
     }
 
     @Test
@@ -122,7 +124,7 @@ class PaneLocatorTest {
                 .parent(100, 101)
                 .parent(101, 100); // cycle, never reaches the pane's pids
         PaneLocator loc = new PaneLocator(pane, parents);
-        assertNull(loc.terminalForPid(100));
+        assertNull(loc.terminalForPid(100).terminal());
     }
 
     @Test
@@ -141,11 +143,43 @@ class PaneLocatorTest {
         };
         HerdrClient noPanes = new FakeHerdr().withNoPanes();
         PaneLocator two = new PaneLocator(noPanes, pane, counting);
-        assertEquals("term_x", two.terminalForPid(7002));
+        assertEquals("term_x", two.terminalForPid(7002).terminal());
         assertEquals(3, calls.get(), "ancestry must be walked once (3 lookups: 7002, 7001, 5000), "
                 + "not re-walked per herdr client");
     }
 
+    // --- fleetd #505: a herdr error during the scan must not read as a clean negative ---------
+
+    @Test
+    void anErrorOnThePaneThatOwnsThePidMakesTheScanIncompleteNotAClearNegative() {
+        // The discriminating case: pane.process_info fails for exactly the pane that DOES own the
+        // caller's pid ("w2:p7", term_a). Before the fix, that failure was swallowed into a plain
+        // "does not own it" and the scan finished with a clean-looking null — indistinguishable
+        // from a real primary. It must now report incomplete, not a definite null.
+        FakeHerdr herdr = new FakeHerdr().processInfoFailsForPane("w2:p7", "transient");
+        PaneLocator loc = new PaneLocator(herdr);
+
+        PaneLocator.Lookup outcome = loc.terminalForPid(FakeHerdr.WORKER_PID);
+
+        assertNull(outcome.terminal(), "the failing pane's ownership could not be confirmed");
+        assertFalse(outcome.complete(),
+                "a scan that could not check the owning pane must not report as complete");
+    }
+
+    @Test
+    void aVanishedPaneThatIsNotTheMatchLeavesAnOtherwiseSuccessfulScanComplete() {
+        // The companion invariant: a DIFFERENT pane (not the caller's own) failing mid-scan must
+        // not turn every mid-scan teardown into a refusal — the real match is still found, and the
+        // scan is still reported complete.
+        FakeHerdr herdr = new FakeHerdr().processInfoFailsForPane("w2:p9", "pane_not_found");
+        PaneLocator loc = new PaneLocator(herdr);
+
+        PaneLocator.Lookup outcome = loc.terminalForPid(FakeHerdr.WORKER_PID);
+
+        assertEquals("term_a", outcome.terminal());
+        assertTrue(outcome.complete(), "a positive match elsewhere in the scan is definitive");
+    }
+
     /** Minimal single-pane {@link HerdrClient} fake, purpose-built for the ancestry tests above. */
     private static final class OnePaneHerdr implements HerdrClient {
         private final ObjectMapper mapper = new ObjectMapper();

fleetd/src/test/java/dev/ltms/fleet/mcp/ConnectionIdentityTest.java

@@ -57,6 +57,23 @@ class ConnectionIdentityTest {
         // must read as "resolved" — the distinction #317 turns on.
         ConnectionIdentity.Caller c = with(_ -> 999_999).resolve("127.0.0.1", 55555);
         assertTrue(c.resolved());
+        assertTrue(c.scanComplete(), "no herdr error happened, so the scan is complete");
+    }
+
+    @Test
+    void scanIsIncompleteWhenHerdrErrorsOnThePaneThatOwnsThePid() {
+        // fleetd #505: a transient herdr error on exactly the pane that DOES own the caller's pid
+        // must be visible as an incomplete scan, distinct from a real primary (resolved(), null
+        // terminal, complete scan). Both have pid > 0 and a null terminal — scanComplete is the
+        // only thing that tells them apart.
+        FakeHerdr failing = new FakeHerdr().processInfoFailsForPane("w2:p7", "transient");
+        ConnectionIdentity id = new ConnectionIdentity(new PaneLocator(failing), _ -> FakeHerdr.WORKER_PID);
+
+        ConnectionIdentity.Caller c = id.resolve("127.0.0.1", 55555);
+
+        assertTrue(c.resolved(), "the pid itself resolved fine — this is not #317's failure");
+        assertNull(c.terminal(), "the owning pane could not be confirmed");
+        assertFalse(c.scanComplete(), "the scan could not check the pane that owns this pid");
     }
 
     @Test

scripts/probe-member-credentials.sh

@@ -65,6 +65,27 @@
 # credential the member holds in full.
 #
 set -uo pipefail
+# `pipefail` is not what catches the parser failure handled below (fleetd #500): in
+# `printf '%s' "$POLICY_JSON" | jq -r '...'`, jq is the LAST element of the pipe, so the pipeline's
+# own exit status is already jq's status, with or without pipefail. It is kept as insurance for if
+# a post-processing stage is ever appended after the parser (e.g. `| tail -n +2`) — at that point
+# the parser would sit upstream and pipefail becomes the only thing that still reports its status.
+
+# --- refuse on an interpreter that cannot run this script (fleetd #500) -------------------------
+#
+# mapfile, used below to parse the policy response, was added in bash 4.0. macOS ships bash 3.2.57
+# at /bin/bash, which predates it. This script's own `set -uo pipefail` does not catch a missing
+# mapfile: the builtin just fails with "command not found" on stderr, and every line below that
+# reads the array it would have filled uses a `:-` default or a slice, neither of which `set -u`
+# catches on an unset array. Left unguarded, that chain ends in the "0 known names" refusal further
+# down — a claim about the POLICY, for a failure that is actually about the INTERPRETER. So the
+# interpreter is checked once, explicitly, before it is asked to do anything mapfile depends on.
+if (( ${BASH_VERSINFO[0]} < 4 )); then
+  echo "refusing to run: this script uses mapfile, which needs bash 4 or newer. This shell is bash" \
+       "${BASH_VERSION:-<unknown, no \$BASH_VERSION>}. Re-run it under a newer bash, for example:" \
+       "\"\$(command -v bash)\" \"$0\"" "$@" >&2
+  exit 3
+fi
 
 FLEETD_HOST="${FLEETD_HOST:-http://127.0.0.1:8765}"
 POLICY_URL="${FLEETD_HOST%/}/member-credentials"
@@ -124,16 +145,32 @@ fi
 # One parse pass: line 1 = present (true/false/null), line 2 = policy mode (possibly blank),
 # lines 3-5 = knownCount/allowedCount/blockedCount, remaining lines = the known[] names. A single
 # pass avoids re-parsing (and re-risking a truthiness bug) five separate times.
+#
+# This used to feed the parser straight into `mapfile -t _FIELDS < <(producer)`. That form cannot
+# see the producer fail: `<` `<(...)` is a process substitution, not a pipeline, so `set -o
+# pipefail` does not reach inside it, and mapfile's own exit status reports whether the BUILTIN
+# ran, not whether the command substituted into it succeeded — a failing jq or python3 there still
+# leaves mapfile at rc=0 with an empty array, read as a parse that genuinely found nothing (fleetd
+# #500). Capturing the parser's output with command substitution first, and checking ITS exit
+# status, reports the producer's real failure while the fact still exists — before it is handed to
+# mapfile at all.
+#
+# mapfile then reads from that captured string with `<<<` (a herestring), not `< <(...)`: `<<<`
+# materialises the whole string in memory first, where `< <(...)` would stream it. That only
+# matters for a large producer; this one is a short credential-name policy response, so the
+# tradeoff is irrelevant here — noted because it would not be for every producer.
 if command -v jq >/dev/null 2>&1; then
-  mapfile -t _FIELDS < <(printf '%s' "$POLICY_JSON" | jq -r '
+  _FIELDS_RAW="$(printf '%s' "$POLICY_JSON" | jq -r '
       (.present | tostring),
       (.policy // ""),
       (.knownCount // 0 | tostring),
       (.allowedCount // 0 | tostring),
       (.blockedCount // 0 | tostring),
-      (.known[]? // empty)')
+      (.known[]? // empty)')"
+  _PARSE_STATUS=$?
+  _PARSER_NAME="jq"
 else
-  mapfile -t _FIELDS < <(printf '%s' "$POLICY_JSON" | python3 - <<'PY'
+  _FIELDS_RAW="$(printf '%s' "$POLICY_JSON" | python3 - <<'PY'
 import json, sys
 data = json.load(sys.stdin)
 print(str(data.get("present")))
@@ -144,7 +181,34 @@ print(data.get("blockedCount") if data.get("blockedCount") is not None else 0)
 for n in (data.get("known") or []):
     print(n)
 PY
-  )
+  )"
+  _PARSE_STATUS=$?
+  _PARSER_NAME="python3"
+fi
+
+if [ "$_PARSE_STATUS" -ne 0 ]; then
+  echo "refusing to run: could not parse the policy fetched from $POLICY_URL — $_PARSER_NAME exited" \
+       "non-zero (status $_PARSE_STATUS). That is a parser failure, not a claim about the policy" \
+       "itself; the policy response has not been read." >&2
+  exit 4
+fi
+
+mapfile -t _FIELDS <<< "$_FIELDS_RAW"
+
+# Arity check — the CORRECTNESS fix (fleetd #500). A parser that exits 0 can still return fewer
+# than the 5 fixed fields (present, policy mode, 3 counts) that every line below this expects,
+# whatever the reason: a producer that printed nothing, malformed JSON that jq/python3 still
+# accepted, or a schema change upstream. The slice just below this (`_FIELDS[@]:5`) does not fire
+# `set -u` on an unset OR a short array, and every fixed-field read above used a `:-` default, so
+# without this check a short `_FIELDS` reaches the "0 known names" guard further down with the
+# same look as a policy that genuinely has 0 names. Check the count here, at the one point the
+# fact is still present, before the slice consumes it.
+if (( ${#_FIELDS[@]} < 5 )); then
+  echo "refusing to run: the policy parser ($_PARSER_NAME) returned ${#_FIELDS[@]} field(s); at" \
+       "least 5 are required (present, policy mode, knownCount, allowedCount, blockedCount). The" \
+       "parse ran but its shape is wrong — this is not a claim about how many names the policy" \
+       "knows." >&2
+  exit 5
 fi
 
 PRESENT="${_FIELDS[0]:-null}"
@@ -164,19 +228,21 @@ case "$KNOWN_COUNT_REPORTED" in
     ;;
 esac
 
-# --- guard the denominator explicitly — never proceed on a zero/short count ---------------------
+# --- guard the denominator explicitly — never proceed on a zero count ---------------------------
 #
-# This is the exact trap named in the ticket: an empty (or truncated) NAMES array passes every
-# subsequent "is it set" check vacuously and prints a table that LOOKS complete. So this is checked
-# before anything else runs, with a message that says why, not just that it failed.
+# This is the exact trap named in the ticket: an empty NAMES array passes every subsequent "is it
+# set" check vacuously and prints a table that LOOKS complete. By this point the interpreter gate,
+# the parser-exit-status check, and the arity check above have already ruled out "the interpreter
+# couldn't run mapfile", "the parser failed", and "the parser returned the wrong shape" — so a zero
+# count reaching here really does mean the policy itself reports 0 known names, not a swallowed
+# failure upstream. That is still checked before anything else runs, with a message that says so.
 if [ "${#NAMES[@]}" -eq 0 ] || [ "$KNOWN_COUNT_REPORTED" -eq 0 ]; then
   cat >&2 <<EOF
 refusing to run: the policy fetched from $POLICY_URL contains 0 known names (present=${PRESENT:-unknown}).
 
-Either memberCredentials: is absent/empty on the running daemon (nothing is protected — see fleetd's
-own startup warning), or the response could not be parsed. Either way, checking zero names would
-print a clean-looking table for a policy that protects nothing, or for a probe that read nothing.
-This is refused rather than reported as a pass.
+memberCredentials: is absent or empty on the running daemon — nothing is protected (see fleetd's own
+startup warning). Checking zero names would print a clean-looking table for a policy that protects
+nothing. This is refused rather than reported as a pass.
 EOF
   exit 1
 fi