Steps to reproduce
- Starts both SCs and one payload.
- Keep 4 sec sleep in avd_process_state_info_queue() at below location:
- TRACE("1. sleep 4s in avd_process_state_info_queue");
- sleep(4);
- TRACE("2. sleep 4s in avd_process_state_info_queue");
+
}
TRACE("queue_size after processing: %lu", (unsigned long) cb->evt_queue.size());
TRACE_LEAVE();
}
- Stop both the controller, compile amfd and replace the binary.
- Start SC-1 followed by SC-2.
- In first node_up message from SC-1, it is sending unassigned alarms from si->update_alarm_state(true); from avd_process_state_info_queue().
- Amfd is resetting the unassigned alarms by calling si->update_alarm_state(false, false); from avd_process_state_info_queue() because of next node_up message from SC-1.
- Duplicates node up messages from SC-1 Amfnd is comming because Amfd is sleeping for 4 sec and Amfd is sending node up messages at an interval of 1 sec. So, Amfnd is sending 4 node up messages to Amfd.
Observed behaviour
The unassigned alarms sent by Amfd after headless resides even if all the SIs are assigned thereafter.
Expected behaviour
The unassigned alarms should be reset after SIs are assigned.
changeset: 7752:4ec13061fdad
branch: opensaf-5.0.x
parent: 7750:8f55b7c1ac4e
user: Nagendra Kumarnagendra.k@oracle.com
date: Mon Jun 27 14:39:18 2016 +0530
summary: amfd: avoid resetting alarm for duplicate node ups [#1893]
changeset: 7753:1619f33dbcbf
tag: tip
parent: 7748:6ef54b502dd4
user: Nagendra Kumarnagendra.k@oracle.com
date: Mon Jun 27 14:39:47 2016 +0530
summary: amfd: avoid resetting alarm for duplicate node ups [#1893]
[staging:4ec130]
[staging:1619f3]
Related
Commit: [1619f3]
Commit: [4ec130]
Tickets:
#1893