Merge pull request #6201 from garlick/issue#6199

include offline nodes in flux overlay errors output

Author: mergify[bot]

doc/man1/flux-overlay.rst

@@ -86,8 +86,9 @@ errors
 
 .. program:: flux overlay errors
 
-:program:`flux overlay errors` summarizes any errors recorded for lost nodes.
-The output consists of one line per unique error with a hostlist prefix.
+:program:`flux overlay errors` summarizes any errors recorded for lost or
+offline nodes.  The output consists of one line per unique error with a
+hostlist prefix.
 
 .. option:: -t, --timeout=FSD
 

src/cmd/builtin/overlay.c

@@ -913,8 +913,11 @@ void gather_errors (flux_t *h,
                 log_msg_exit ("error adding to error hash");
         }
         else if (streq (status, "offline")) {
-            /* Don't report offline nodes.
-             */
+            // report offline only if there is error text
+            if (error && strlen (error) > 0) {
+                if (errhash_add_one (errhash, child_rank, error) < 0)
+                    log_msg_exit ("error adding to error hash");
+            }
         }
         else { // recurse
             gather_errors (h, child_rank, errhash, timeout);

t/t3303-system-healthcheck.t

@@ -139,9 +139,9 @@ test_expect_success 'flux overlay status shows rank 3 offline' '
 	test_cmp health.exp health.out
 '
 
-test_expect_success 'flux overlay errors prints nothing' '
+test_expect_success 'flux overlay errors shows rank 3' '
 	flux overlay errors --timeout=0 >errors2.out &&
-	test $(wc -l <errors2.out) -eq 0
+	grep "fake3: administrative shutdown" errors2.out
 '
 
 test_expect_success 'flux overlay status --summary' '
@@ -202,9 +202,8 @@ test_expect_success 'ping to rank 14 fails with EHOSTUNREACH' '
 '
 
 test_expect_success 'flux overlay errors shows the lost connection' '
-	echo "fake14: lost connection" >errors3.exp &&
 	flux overlay errors --timeout=0 >errors3.out &&
-	test_cmp errors3.exp errors3.out
+	grep "fake14: lost connection" errors3.out
 '
 
 test_expect_success 'wait for rank 0 subtree to be degraded' '