Changeset 10661 for trunk/Ohana/src/opihi/pcontrol/StartJob.c

Timestamp:

Dec 12, 2006, 10:45:22 AM (20 years ago)

Author:

eugene

Message:

better solution for hung hosts

File:

• trunk/Ohana/src/opihi/pcontrol/StartJob.c (modified) (1 diff)

trunk/Ohana/src/opihi/pcontrol/StartJob.c

@@ -40 +40 @@
       // we need the job start to return a valid Job ID, 
       // give up on jobs which don't get started.
+      // XXX we are sensitive here to the time it takes pclient
+      // to fork the job.  if this is slow, the client may appear to hang.
       gprint (GP_ERR, "host %s is not responding (StartJob)\n", host[0].hostname);
       if (VerboseMode()) gprint (GP_ERR, "host %s is not responding\n", host[0].hostname);
-      goto failure;
+
+      // unlink host & job
+      job[0].host = NULL;
+      host[0].job = NULL;
+      PutHost (host, PCONTROL_HOST_DONE, STACK_BOTTOM);
+      PutJob (job, PCONTROL_JOB_PENDING, STACK_BOTTOM);
+      FreeIOBuffer (&buffer);
+      return (FALSE);
 
     case PCLIENT_GOOD: