cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements

223
Views
1
Helpful
7
Replies
khgrant
Cisco Employee

SNMP trap for HA

 

Hi team,

 

 

I have a question regarding SNMP trap.

 

Can NSO system send the SNMP trap at the HA slave failure timing?

 

In my test when the Master fails (down), SNMP trap will be sent by new master NSO.

 

However when the Slave NSO fails, no SNMP trap will be sent.

 

Is there any way/config to enable the SNMP trap at the Slave NSO failure?

 

 

Thanks

 

 

Gaku

 

7 REPLIES 7
khgrant
Cisco Employee

 

If it fails, then it probably  (or at least possibly)  isn’t able to do anything.

 

Cheers,

 

KJ.

 

khgrant
Cisco Employee

 

Hi KJ,

 

 

Thanks for your comment.

 

So for in case of Slave Failure, should we track the Slave NSO by some other system?

 

 

 

Gaku

 

khgrant
Cisco Employee

 

 

I'm confused here.

 

 

Which traps do you see. AFAIK no traps are ever sent by NSO HA code.

 

 

NSO generates SNMP traps when NSO alarms are created. The list of alarm types can be found under:

 

 

$NCS_DIR/src/ncs/yang/tailf-ncs-alarms.yang

 

 

/klacke

 

khgrant
Cisco Employee

 

 

Ahh, beautiful. So 3 different alarms in tailf-hcc-alarms.yang

 

 

   identity node-failure {

 

     base hcc-alarm;

 

     description

 

       "The node lost HA connection with its master";

 

   }

 

 

   identity device-node-failure {

 

     base hcc-alarm;

 

     description

 

       "A service node noticed one of its device nodes lost

 

        HA connection with its master";

 

   }

 

 

   identity bgp-failure {

 

     base hcc-alarm;

 

     description

 

       "A service node noticed it lost its BGP prefix path to

 

        its master";

 

   }

 

khgrant
Cisco Employee

 

Hi Klacke,

 

 

Those 2 alarms(identity node-failure, identity device-node-failure) are sent by New-Master (original slave), when their original master is gone, correct?

 

khgrant
Cisco Employee

 

Checking the code:

 

 

I see one alarm getting created when there has been a fail over. I.e when a slave goes to master mode.

 

Type: node-failure

 

Text:

 

 

  "HA connection lost. '" + localNode +

 

   "' transitioning to HA MASTER role. " +

 

   "When the problem has been fixed, "+

 

"role-override the old MASTER to SLAVE to "+

 

   "prevent config loss, then role-revert all nodes. "+

 

   "This will clear the alarm.";

 

 

 

The other one device-node-failure you can  ignore unless you're using clustering and ha device nodes which is .. well ... very complicated and involved.

 

 

/klacke

 

khgrant
Cisco Employee

 

Hi Klacke,

 

 

Thank you very much!!

 

 

Gaku

 

Create
Recognize Your Peers
Content for Community-Ad