Merge pull request flux-framework#4131 from garlick/overlay_errstr

flux-overlay: fix timeout error message
chu11 · Feb 15, 2022 · a0bfb7f · a0bfb7f
2 parents 6f0165b + d6e2aee
commit a0bfb7f
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 2 deletions.
diff --git a/src/cmd/builtin/overlay.c b/src/cmd/builtin/overlay.c
@@ -425,6 +425,7 @@ static int status_healthwalk (struct status *ctx,
     struct status_node node = { .ghost = false, .connector = connector };
     flux_future_t *f;
     json_t *children;
+    const char *errstr;
     int rc = 0;
 
     monotime (&ctx->start);
@@ -436,19 +437,20 @@ static int status_healthwalk (struct status *ctx,
                                 "status", &node.status,
                                 "duration", &node.duration,
                                 "children", &children) < 0) {
+        errstr = future_strerror (f, errno);
         /* RPC failed.
          * An error at level 0 should be fatal, e.g. unknown wait argument,
          * bad rank, timeout.  An error at level > 0 should return -1 so
          * ghostwalk() can be tried (parent hasn't noticed child crash?)
          * and sibling subtrees can be probed.
          */
         if (level == 0)
-            log_msg_exit ("%s", future_strerror (f, errno));
+            log_msg_exit ("%s", errstr);
         printf ("%s%s%s: %s%s\n",
                 status_indent (ctx, level),
                 connector_string (connector),
                 status_getname (ctx, rank),
-                future_strerror (f, errno),
+                errstr,
                 status_rpctime (ctx));
         rc = -1;
         goto done;

diff --git a/t/t3303-system-healthcheck.t b/t/t3303-system-healthcheck.t
@@ -229,4 +229,17 @@ test_expect_success 'flux overlay disconnect fails on bad input' '
 	grep "TARGET must be a valid rank or hostname" disconn3.err
 '
 
+test_expect_success 'stop broker 12' '
+	$startctl kill 12 19
+'
+
+test_expect_success 'flux overlay status prints connection timed out on 12' '
+	flux overlay status --no-pretty --no-color >status.out &&
+	grep "fake12: Connection timed out" status.out
+'
+
+test_expect_success 'continue broker 12' '
+	$startctl kill 12 18
+'
+
 test_done