Hi Anders,

Ack from me with a minor comment.
Shouldn't the log message say "... Sync was killed ..." instead of "... Sync 
was aborted ..." ?

Best regards,
Zoran

-----Original Message-----
From: Anders Björnerstedt 
Sent: Thursday, April 09, 2015 7:04 AM
To: [email protected]; Zoran Milinkovic
Cc: [email protected]
Subject: [PATCH 1 of 1] IMM: Halt sync with SIGKILL insted of SIGTERM and 
replace assert with error [#1295]

 osaf/services/saf/immsv/immnd/ImmModel.cc |  6 +++++-  
osaf/services/saf/immsv/immnd/immnd_evt.c |  2 +-
 2 files changed, 6 insertions(+), 2 deletions(-)


The problem was triggerde by the handling of the unusual case of a PRTA update 
failing during on going sync due to PBE being configured but unavailable at 
that time. Such a failed PRTA update has to abort the sync becasue the PRTA 
update may succeed at sync clients yet fail at veterans.
This is due to the sync clients possibly not yet having synced the config data 
specifying if PBE is available or not.

The handling of this case by the IMMND coord was to send a SIGTERM to the 
sync-process and to broacast an abort of the cluster wide sync job over fevs.
A SIGERM may however be slow to get processed by the sync process. If the abort 
of the sync job gets handled before the sync process has terminated; and the 
sync process sends additional sync messages arriving at a processor after the 
sync has been aborted, then the assert will get triggered.

The solution is to replace the SIGTERM with SIGKILL. THat should drastically 
reduce the probability of the sync process lingering after the sync job has 
been aborted in the IMMND cluster. In addition the assert is dissarmed and 
replaced with a return of error (to the sync process) for this case.

diff --git a/osaf/services/saf/immsv/immnd/ImmModel.cc 
b/osaf/services/saf/immsv/immnd/ImmModel.cc
--- a/osaf/services/saf/immsv/immnd/ImmModel.cc
+++ b/osaf/services/saf/immsv/immnd/ImmModel.cc
@@ -10892,7 +10892,11 @@ SaAisErrorT ImmModel::nextSyncResult(Imm
 
     osafassert(!(obj->mObjFlags & IMM_CREATE_LOCK));
     osafassert(!(obj->mObjFlags & IMM_DELETE_LOCK));
-    osafassert(!(obj->mObjFlags & IMM_RT_UPDATE_LOCK)); /* Is this really 
secured? */
+    if(obj->mObjFlags & IMM_RT_UPDATE_LOCK) {
+        LOG_NO("ERR_NO_RESOURCES: Sync was aborted due to interference with 
blocked PRTA update");
+        err = SA_AIS_ERR_NO_RESOURCES;
+        goto done;
+    }
 
     /* (1) op.addObject(objectName);*/
 
diff --git a/osaf/services/saf/immsv/immnd/immnd_evt.c 
b/osaf/services/saf/immsv/immnd/immnd_evt.c
--- a/osaf/services/saf/immsv/immnd/immnd_evt.c
+++ b/osaf/services/saf/immsv/immnd/immnd_evt.c
@@ -6380,7 +6380,7 @@ static void immnd_evt_proc_rt_object_mod
                                immnd_abortSync(cb);
                                if(cb->syncPid > 0) {
                                        LOG_WA("STOPPING sync process pid %u", 
cb->syncPid);
-                                       kill(cb->syncPid, SIGTERM);
+                                       kill(cb->syncPid, SIGKILL); /* 
Immediate stop of the sync process. 
+*/
                                }
                        }
                }

------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
_______________________________________________
Opensaf-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-devel

Reply via email to