slingshot_metrics: don't return an error from sample() when a device'…

…s counter lookup fails
ovis-hpc · Nov 24, 2024 · fb97e1c · fb97e1c
1 parent 317058e
commit fb97e1c
Showing 1 changed file with 3 additions and 1 deletion.
diff --git a/ldms/src/sampler/slingshot_metrics/slingshot_metrics.c b/ldms/src/sampler/slingshot_metrics/slingshot_metrics.c
@@ -593,6 +593,8 @@ static int sample(struct ldmsd_sampler *self)
                                             &counters.value[j],
                                             NULL);
                         if (rc != 0) {
+                                ovis_log(mylog, OVIS_LWARNING, "sample(): cxil_read_cntr() failed for device %s\n",
+                                         dev_info->device_name);
                                 cache_cxil_dev_close(dev);
                                 /* FIXME - we should really free the record here,
                                    and avoid adding it to the list, but there is
@@ -606,7 +608,7 @@ static int sample(struct ldmsd_sampler *self)
         }
         base_sample_end(sampler_base);
 
-        return rc;
+        return 0;
 }
 
 static void term(struct ldmsd_plugin *self)