[Darshan-commits] [Git][darshan/darshan][master] soften error handling in lustre module

[Philip Carns]

Philip Carns pushed to branch master at darshan / darshan


Commits:
52c1c54a by Phil Carns at 2018-10-03T17:45:58Z
soften error handling in lustre module

Fixes #243

- - - - -


1 changed file:

- darshan-runtime/lib/darshan-lustre.c


Changes:

=====================================
darshan-runtime/lib/darshan-lustre.c
=====================================
--- a/darshan-runtime/lib/darshan-lustre.c
+++ b/darshan-runtime/lib/darshan-lustre.c
@@ -242,9 +242,16 @@ static void lustre_shutdown(
         {
             rec_ref = darshan_lookup_record_ref(lustre_runtime->record_id_hash,
                 &shared_recs[i], sizeof(darshan_record_id));
-            assert(rec_ref);
-
-            rec_ref->record->base_rec.rank = -1;
+            /* As in other modules, it should not be possible to lose a
+             * record after we have already performed a collective to
+             * identify that it is shared with other ranks.  We print an
+             * error msg and continue rather than asserting in this case,
+             * though, see #243.
+             */
+            if(rec_ref)
+                rec_ref->record->base_rec.rank = -1;
+            else
+                fprintf(stderr, "WARNING: unexpected condition in Darshan, possibly triggered by memory corruption.  Darshan log may be incorrect.\n");
         }
 
         /* sort the array of files descending by rank so that we get all of the 



View it on GitLab: https://xgitlab.cels.anl.gov/darshan/darshan/commit/52c1c54a3ce33e423077903dd10d8304b9cf6abe

-- 
View it on GitLab: https://xgitlab.cels.anl.gov/darshan/darshan/commit/52c1c54a3ce33e423077903dd10d8304b9cf6abe
You're receiving this email because of your account on xgitlab.cels.anl.gov.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/darshan-commits/attachments/20181003/24f2b651/attachment.html>