Changeset 18 for trunk

Timestamp:

05/31/11 17:23:32 (14 years ago)

Author:

mmamonski

Message:

log reader race condition causing hanging of application

File:

• trunk/pbs_drmaa/log_reader.c (modified) (2 diffs)

trunk/pbs_drmaa/log_reader.c

@@ -424 +424 @@
                                                         {
                                                                 fsd_log_debug(("WT - forcing update of job: %s", temp_job->job_id ));
-                                                                temp_job->update_status( temp_job );
+                                                                TRY
+                                                                {
+                                                                        temp_job->update_status( temp_job );
+                                                                }
+                                                                EXCEPT_DEFAULT
+                                                                {
+                                                                        /*TODO: distinguish between invalid job and internal errors */
+                                                                        fsd_log_debug(("Job finished just after entering running state: %s", temp_job->job_id));
+                                                                }
+                                                                END_TRY
                                                         }
                                                         else
@@ -503 +512 @@
                  {
                         const fsd_exc_t *e = fsd_exc_get();
-
-                        fsd_log_error(( "%s: <%d:%s>", self->name, e->code(e), e->message(e) ));
-                        fsd_exc_reraise();
+                        /* Its better to exit and communicate error rather then let the application to hang */
+                        fsd_log_fatal(( "Exception in wait thread %s: <%d:%s>. Exiting !!!", self->name, e->code(e), e->message(e) ));
+                        exit(1);
                  }
                 END_TRY