Hi Gary,
The node up messages is invalid from Mds point of view and so let's not
process the node up message itself. So, keeping the NODE_UP check as first
check in the function is better option.
Thanks
-Nagu
> -----Original Message-----
> From: Gary Lee [mailto:[email protected]]
> Sent: 25 October 2016 13:50
> To: Nagendra Kumar
> Cc: [email protected]; Praveen Malviya;
> [email protected]; [email protected]; opensaf-
> [email protected]
> Subject: Re: [PATCH 1 of 1] amfd: ignore node_up until the mds event amfnd
> up has been received [#2124]
>
> Hi Nagu
>
> That if statement could be anywhere before messages to amfnd are sent. It's
> just that the headless recovery node counting code can still continue to run
> as it's not sending anything to amfnd. So I think we can put the new if
> statement after this?
>
> thanks
> Gary
>
> > On 25 Oct. 2016, at 5:34 pm, Nagendra Kumar <[email protected]>
> wrote:
> >
> > Yes Gary. I wanted it to be first check in the function. You see problem
> > with
> it?
> >
> > Thanks
> > -Nagu
> >> -----Original Message-----
> >> From: Gary Lee [mailto:[email protected]]
> >> Sent: 25 October 2016 11:49
> >> To: Nagendra Kumar
> >> Cc: [email protected]; Praveen Malviya;
> >> [email protected]; [email protected]; opensaf-
> >> [email protected]
> >> Subject: Re: [PATCH 1 of 1] amfd: ignore node_up until the mds event
> >> amfnd up has been received [#2124]
> >>
> >> Hi Nagu
> >>
> >> "Hi Gary,
> >>
> >> You can move the following changes in the beginning of the function
> >> avd_node_up_evh():
> >>
> >> + if (amfnd_svc_db->find(n2d_msg->msg_info.n2d_node_up.node_id)
> >> == amfnd_svc_db->end()) {
> >> + // don't process node_up until svc up is received
> >> + LOG_WA("amfnd svc up not yet received from node %x",
> >> n2d_msg->msg_info.n2d_node_up.node_id);
> >> + goto done;
> >> + }
> >> +”
> >>
> >> I thought you were suggesting to move the if statement to higher up
> >> in the function?
> >>
> >> Thanks
> >> Gary
> >>
> >>
> >>> On 25 Oct. 2016, at 5:17 pm, Nagendra Kumar
> <[email protected]>
> >> wrote:
> >>>
> >>> You mean to store SUSI messages from Amfnd ?
> >>>
> >>> Thanks
> >>> -Nagu
> >>>> -----Original Message-----
> >>>> From: Gary Lee [mailto:[email protected]]
> >>>> Sent: 25 October 2016 11:44
> >>>> To: Nagendra Kumar
> >>>> Cc: [email protected]; Praveen Malviya;
> >>>> [email protected]; [email protected];
> opensaf-
> >>>> [email protected]
> >>>> Subject: Re: [PATCH 1 of 1] amfd: ignore node_up until the mds
> >>>> event amfnd up has been received [#2124]
> >>>>
> >>>> Hi Nagu
> >>>>
> >>>> I did move it up higher in avd_node_up_evh() as you suggest. But I
> >>>> think we should leave it after the headless logic.
> >>>>
> >>>> Thanks
> >>>> Gary
> >>>>
> >>>>> On 25 Oct. 2016, at 5:08 pm, Nagendra Kumar
> >> <[email protected]>
> >>>> wrote:
> >>>>>
> >>>>> Hi Gary,
> >>>>> The following changes (line no. 420) are the same as in previous
> >>>>> email,
> >>>> please change:
> >>>>>
> >>>>> + if (amfnd_svc_db->find(n2d_msg-
> >msg_info.n2d_node_up.node_id)
> >>>> == amfnd_svc_db->end()) {
> >>>>> + // don't process node_up until svc up is received
> >>>>> + LOG_NO("amfnd svc up not yet received from node %x",
> >>>> n2d_msg->msg_info.n2d_node_up.node_id);
> >>>>> + goto done;
> >>>>> + }
> >>>>> +
> >>>>>
> >>>>> Thanks
> >>>>> -Nagu
> >>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: Gary Lee [mailto:[email protected]]
> >>>>>> Sent: 25 October 2016 07:23
> >>>>>> To: [email protected]; Praveen Malviya; Nagendra
> Kumar;
> >>>>>> [email protected]; [email protected]
> >>>>>> Cc: [email protected]
> >>>>>> Subject: [PATCH 1 of 1] amfd: ignore node_up until the mds event
> >>>>>> amfnd up has been received [#2124]
> >>>>>>
> >>>>>> osaf/services/saf/amf/amfd/include/node.h | 1 +
> >>>>>> osaf/services/saf/amf/amfd/main.cc | 1 +
> >>>>>> osaf/services/saf/amf/amfd/mds.cc | 18 ++++++++++--------
> >>>>>> osaf/services/saf/amf/amfd/ndfsm.cc | 20
> ++++++++++++++++++--
> >>>>>> 4 files changed, 30 insertions(+), 10 deletions(-)
> >>>>>>
> >>>>>>
> >>>>>> if the svc up event for amfnd arrives after N2D_NODE_UP, then
> >>>>>> amfd may fail to send D2N_NODE_UP to amfnd. This will eventually
> >>>>>> cause the respective amfnd to reboot the node, due to message ID
> >> mismatches.
> >>>>>> Instead, ignore node up until we have received the mds event
> >>>>>> 'amfnd up' for that node.
> >>>>>>
> >>>>>> diff --git a/osaf/services/saf/amf/amfd/include/node.h
> >>>>>> b/osaf/services/saf/amf/amfd/include/node.h
> >>>>>> --- a/osaf/services/saf/amf/amfd/include/node.h
> >>>>>> +++ b/osaf/services/saf/amf/amfd/include/node.h
> >>>>>> @@ -162,6 +162,7 @@ struct NodeNameCompare: public std::bina
> >>>> extern
> >>>>>> AmfDb<std::string, AVD_AVND> *node_name_db; extern
> >>>> AmfDb<uint32_t,
> >>>>>> AVD_AVND> *node_id_db; extern AmfDb<uint32_t,
> >>>> AVD_FAIL_OVER_NODE>
> >>>>>> *node_list_db;
> >>>>>> +extern std::set<uint32_t> *amfnd_svc_db;
> >>>>>> extern std::map<SaClmNodeIdT, MDS_SVC_PVT_SUB_PART_VER>
> >>>>>> nds_mds_ver_db; class AVD_AMF_NG {
> >>>>>> public:
> >>>>>> diff --git a/osaf/services/saf/amf/amfd/main.cc
> >>>>>> b/osaf/services/saf/amf/amfd/main.cc
> >>>>>> --- a/osaf/services/saf/amf/amfd/main.cc
> >>>>>> +++ b/osaf/services/saf/amf/amfd/main.cc
> >>>>>> @@ -573,6 +573,7 @@ static uint32_t initialize(void)
> >>>>>> base::GetEnv("OSAF_AMF_MIN_CLUSTER_SIZE", uint32_t{2});
> >>>>>>
> >>>>>> node_list_db = new AmfDb<uint32_t, AVD_FAIL_OVER_NODE>;
> >>>>>> + amfnd_svc_db = new std::set<uint32_t>;
> >>>>>> /* get the node id of the node on which the AVD is running. */
> >>>>>> cb->node_id_avd = m_NCS_GET_NODE_ID;
> >>>>>>
> >>>>>> diff --git a/osaf/services/saf/amf/amfd/mds.cc
> >>>>>> b/osaf/services/saf/amf/amfd/mds.cc
> >>>>>> --- a/osaf/services/saf/amf/amfd/mds.cc
> >>>>>> +++ b/osaf/services/saf/amf/amfd/mds.cc
> >>>>>> @@ -420,16 +420,18 @@ static uint32_t
> avd_mds_svc_evt(MDS_CALL
> >>>>>>
> >>>>>> case NCSMDS_SVC_ID_AVND:
> >>>>>> {
> >>>>>> + AVD_EVT *evt = new AVD_EVT();
> >>>>>> +
> >>>>>> + evt->rcv_evt = AVD_EVT_MDS_AVND_UP;
> >>>>>> + evt->info.node_id =
> >>>>>> m_NCS_NODE_ID_FROM_MDS_DEST(evt_info->i_dest);
> >>>>>> if (evt_info->i_node_id == cb->node_id_avd)
> >>>> {
> >>>>>> - AVD_EVT *evt = new AVD_EVT();
> >>>>>> + cb->local_avnd_adest = evt_info-
> >>>>>>> i_dest;
> >>>>>> + }
> >>>>>> + if (m_NCS_IPC_SEND(&cb->avd_mbx, evt,
> >>>>>> NCS_IPC_PRIORITY_HIGH) != NCSCC_RC_SUCCESS) {
> >>>>>> + LOG_ER("%s: ncs_ipc_send failed",
> >>>>>> __FUNCTION__);
> >>>>>> + delete evt;
> >>>>>> + }
> >>>>>>
> >>>>>> - evt->rcv_evt =
> >>>>>> AVD_EVT_MDS_AVND_UP;
> >>>>>> - cb->local_avnd_adest = evt_info-
> >>>>>>> i_dest;
> >>>>>> - if (m_NCS_IPC_SEND(&cb-
> >>>>>>> avd_mbx, evt, NCS_IPC_PRIORITY_HIGH) != NCSCC_RC_SUCCESS) {
> >>>>>> - LOG_ER("%s: ncs_ipc_send
> >>>>>> failed", __FUNCTION__);
> >>>>>> - delete evt;
> >>>>>> - }
> >>>>>> - }
> >>>>>> //Post MDS version info to mailbox.
> >>>>>> AVD_EVT *evt1 = new AVD_EVT();
> >>>>>> evt1->rcv_evt =
> >>>>>> AVD_EVT_ND_MDS_VER_INFO;
> >>>>>> diff --git a/osaf/services/saf/amf/amfd/ndfsm.cc
> >>>>>> b/osaf/services/saf/amf/amfd/ndfsm.cc
> >>>>>> --- a/osaf/services/saf/amf/amfd/ndfsm.cc
> >>>>>> +++ b/osaf/services/saf/amf/amfd/ndfsm.cc
> >>>>>> @@ -32,6 +32,10 @@
> >>>>>>
> >>>>>> AmfDb<uint32_t, AVD_FAIL_OVER_NODE> *node_list_db = 0; /*
> >>>>>> SaClmNodeIdT index */
> >>>>>>
> >>>>>> +// indicates whether MDS service up has been received from amfnd.
> >>>>>> +// If a node ID is in the set, then service up has been received.
> >>>>>> +std::set<uint32_t> *amfnd_svc_db = 0;
> >>>>>> +
> >>>>>>
> >>>>>>
> >>>>
> >>
> /*************************************************************
> >>>>>> ****************
> >>>>>> * Function: avd_process_state_info_queue
> >>>>>> *
> >>>>>> @@ -340,6 +344,12 @@ void avd_node_up_evh(AVD_CL_CB *cb,
> >> AVD_
> >>>>>> }
> >>>>>> }
> >>>>>>
> >>>>>> + if
> >>>>>> + (amfnd_svc_db->find(n2d_msg->msg_info.n2d_node_up.node_id)
> >>>>>> == amfnd_svc_db->end()) {
> >>>>>> + // don't process node_up until svc up is received
> >>>>>> + LOG_NO("amfnd svc up not yet received from node %x",
> >>>>>> n2d_msg->msg_info.n2d_node_up.node_id);
> >>>>>> + goto done;
> >>>>>> + }
> >>>>>> +
> >>>>>> /* Cannot use avd_msg_sanity_chk here since this is a special case
> */
> >>>>>> if ((avnd = avd_node_find_nodeid(n2d_msg-
> >>>>>>> msg_info.n2d_node_up.node_id)) == nullptr) {
> >>>>>> TRACE("invalid node ID (%x)", n2d_msg-
> >>>>>>> msg_info.n2d_node_up.node_id);
> >>>>>> @@ -645,8 +655,13 @@ void avd_nd_ncs_su_failed(AVD_CL_CB
> *cb,
> >>>>>>
> >>>>>> void avd_mds_avnd_up_evh(AVD_CL_CB *cb, AVD_EVT *evt) {
> >>>>>> - TRACE("Local node director is up, start sending heart beats to %"
> >>>>>> PRIx64, cb->local_avnd_adest);
> >>>>>> - avd_tmr_snd_hb_evh(cb, evt);
> >>>>>> + if (evt->info.node_id == cb->node_id_avd) {
> >>>>>> + TRACE("Local node director is up, start sending heart
> >>>>>> + beats
> >>>>>> to %" PRIx64, cb->local_avnd_adest);
> >>>>>> + avd_tmr_snd_hb_evh(cb, evt);
> >>>>>> + }
> >>>>>> +
> >>>>>> + TRACE("amfnd on %x is up", evt->info.node_id);
> >>>>>> + amfnd_svc_db->insert(evt->info.node_id);
> >>>>>> }
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>
> >>
> /*************************************************************
> >>>>>> ****************
> >>>>>> @@ -672,6 +687,7 @@ void avd_mds_avnd_down_evh(AVD_CL_CB
> >> *cb
> >>>>>>
> >>>>>> //update MDS version db.
> >>>>>> nds_mds_ver_db.erase(evt->info.node_id);
> >>>>>> + amfnd_svc_db->erase(evt->info.node_id);
> >>>>>>
> >>>>>> if (node != nullptr) {
> >>>>>> // Do nothing if the local node goes down. Most likely due
> >>>>>> to system shutdown.
> >>>>
> >>
>
------------------------------------------------------------------------------
The Command Line: Reinvented for Modern Developers
Did the resurgence of CLI tooling catch you by surprise?
Reconnect with the command line and become more productive.
Learn the new .NET and ASP.NET CLI. Get your free copy!
http://sdm.link/telerik
_______________________________________________
Opensaf-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
