#3250 imm: immnd dectect headless state wrong

[Thien Minh Huynh]

from MDS thread, that receive in order event : IMMD RED_DOWN from standby , NCSMDS_DOWN, IMMD RED_DOWN from active. All event from MDS thread will be going to IMMND mail box. In event NCSMDS_DOWN was mark active node down.

IMMND receive IMMD RED_DOWN (standby) from MDS thread, then enter the headless state. IMMND enter the headless state due to ACTIVE and STANDBY is down. So the entering headless state is unexpected.

<143>1 2021-01-25T21:49:35.467747+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="115"] 4916:imm/common/immsv_evt.c:6569 T8 Received: IMMND_EVT_MDS_INFO (1) from 0
<143>1 2021-01-25T21:49:35.467758+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="116"] 4916:imm/immnd/immnd_evt.c:12290 T2 IMMD RED_DOWN EVENT 2010f role=2 ==> ACT:0 SBY:0
<140>1 2021-01-25T21:49:35.468554+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="117"] 4916:imm/immnd/immnd_evt.c:12292 WA Both Active & Standby DOWN, going to headless
<140>1 2021-01-25T21:49:35.468578+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="118"] 4916:imm/immnd/immnd_evt.c:12315 WA SC Absence IS allowed:900 IMMD service is DOWN ex_immd_node_id=2020f
<141>1 2021-01-25T21:49:35.468595+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="119"] 4916:imm/immnd/immnd_evt.c:12395 NO IMMD SERVICE IS DOWN, HYDRA IS CONFIGURED => UNREGISTERING IMMND form MDS
<143>1 2021-01-25T21:49:35.468602+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="120"] 4916:imm/immnd/immnd_mds.c:256 >> immnd_mds_unregister
...
<143>1 2021-01-25T21:49:36.495445+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="237"] 4916:imm/common/immsv_evt.c:6569 T8 Received: IMMND_EVT_MDS_INFO (1) from 0
<143>1 2021-01-25T21:49:36.49548+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="238"] 4916:imm/immnd/immnd_evt.c:12252 T2 NCSMDS_DOWN 0
<140>1 2021-01-25T21:49:36.495548+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="239"] 4916:imm/immnd/immnd_evt.c:12315 WA SC Absence IS allowed:900 IMMD service is DOWN ex_immd_node_id=2020f
<141>1 2021-01-25T21:49:36.495582+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="240"] 4916:imm/immnd/immnd_evt.c:12395 NO IMMD SERVICE IS DOWN, HYDRA IS CONFIGURED => UNREGISTERING IMMND form MDS
<143>1 2021-01-25T21:49:36.495595+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="241"] 4916:imm/immnd/immnd_mds.c:256 >> immnd_mds_unregister
...
<143>1 2021-01-25T21:49:37.49666+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="275"] 4916:imm/common/immsv_evt.c:6569 T8 Received: IMMND_EVT_MDS_INFO (1) from 0
<143>1 2021-01-25T21:49:37.496675+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="276"] 4916:imm/immnd/immnd_evt.c:12290 T2 IMMD RED_DOWN EVENT 2020f role=1 ==> ACT:0 SBY:0
<140>1 2021-01-25T21:49:37.496744+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="277"] 4916:imm/immnd/immnd_evt.c:12292 WA Both Active & Standby DOWN, going to headless
<140>1 2021-01-25T21:49:37.496786+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="278"] 4916:imm/immnd/immnd_evt.c:12315 WA SC Absence IS allowed:900 IMMD service is DOWN ex_immd_node_id=2020f
<141>1 2021-01-25T21:49:37.496809+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="279"] 4916:imm/immnd/immnd_evt.c:12395 NO IMMD SERVICE IS DOWN, HYDRA IS CONFIGURED => UNREGISTERING IMMND form MDS
<143>1 2021-01-25T21:49:37.496819+11:00 PL-2-4 osafimmnd 4916 osafimmnd [meta sequenceId="280"] 4916:imm/immnd/immnd_mds.c:256 >> immnd_mds_unregister

[Thien Minh Huynh]

commit 54bb248ee217f5ba619a7c2389cb10e2b0fefaab (HEAD -> develop, origin/develop)
Author: thien.m.huynh thien.m.huynh@dektech.com.au
Date: Thu Jan 28 15:58:19 2021 +0700

imm: fix immnd detect headless state wrong [#3250]

From MDS thread, that receives in order event: IMMD_RED_DOWN from standby,
NCSMDS_DOWN, IMMD_RED_DOWN from active. All event from MDS thread will be
sent to the mail box. MDS thread marks active down when receive NCSMDS_DOWN
event. In main thread the IMMND enter headless state when receive the first
event IMMD_RED_DOWN. Headless state is unexpected.

The solution just mark node down when receive NCSMDS_DOWN on the main thread.
That is prevent IMMND detect headless state wrong.

[Thien Minh Huynh]

commit 76b7a650effce95575add26d5b25d500b5a441fd (HEAD -> develop, origin/develop)
Author: thien.m.huynh thien.m.huynh@dektech.com.au
Date: Fri Feb 5 13:26:57 2021 +0700

imm: ignore enter headless if already entered [#3250]

immnd going to headless twice when receive event NCSMDS_DOWN, NCSMDS_RED_DOWN
in the same time.

The fix is prevent go to headless many time.