SNMP Trap Reciver

[Armen Martirosyan]

Gentlemen,

Can you please explain what is the SNMP trap receiver's function? I am generating some traps from a switch but don't see it recorded anywhere.

 

Thank you

 

[Sunil Rawoorkar]

Hi Armen,

SNMP trap is used staged ZTP process, In the bootstrap stage,during leaf device boot-strap it gets generic config and then openclos find the topology and applies the new topology.   
Please see "openclos.yaml"  on how to set ZTP to 2-stage,  set SNMP trap settings and other details.

Thanks,
Sunil

[IBM Client Innovation Center - Networking Services]

Hi,

With OpenClos-devR3.0:

I face a similar behavior where the leaves (qfx5200-32c-32q, 15.1X53-D30.5) have been junos upgraded and have received the 1st stage configuration (all leaf devices name is "netboot" and basic configuration without BGP)
Then I see traffic from leaf devices to the openclos server on the trapd port 20162. But the openclos-trapd.log does not point to anything related to the 2nd stage where the final leaf device configuration is generated and uploaded.

Could you please provide your recommendation about the next steps to tbshoot and understand why the 2nd stage does not start for the leaf devices.
For information, the spine devices are QFX10002 and they got the final configuration (2nd stage completed).

Many thanks for your support.
Regards, Ludovic

[IBM Client Innovation Center - Networking Services]

Hi,

In addition, I have set debug level for trapd in conf/logging.yaml and also executed trapd.py in addition to sampleApplication.
I see now in the openclos-trapd.log the following lines regulary for both leaves 192.168.117.75 and 192.168.117.76:
--
...
2016-11-22 16:45:35,981 [devicePlugin] [DEBUG   ] [7f8109ffb700] Checking all pods for ip 192.168.117.75
2016-11-22 16:45:35,984 [devicePlugin] [DEBUG   ] [7f8109ffb700] Checking pod[id='2913c995-f8c8-495b-9be9-6e8eefc1ce4c', name='labLeafSpine']: 192.168.117.71/24
2016-11-22 16:45:35,986 [devicePlugin] [ERROR   ] [7f8109ffb700] Couldn't find any pod containing device ip: 192.168.117.75
...
2016-11-22 16:43:37,437 [devicePlugin] [DEBUG   ] [7f81219c8700] Checking all pods for ip 192.168.117.76
2016-11-22 16:43:37,440 [devicePlugin] [DEBUG   ] [7f81219c8700] Checking pod[id='2913c995-f8c8-495b-9be9-6e8eefc1ce4c', name='labLeafSpine']: 192.168.117.71/24
2016-11-22 16:43:37,442 [devicePlugin] [ERROR   ] [7f81219c8700] Couldn't find any pod containing device ip: 192.168.117.76
...

The fact that these IP addresses (obtained via openclos dhcpd) are not known in openclos pod DB can explain why teh 2nd stage does not start.

Would you have any advice to resolve this?

Thanks, Ludovic

[IBM Client Innovation Center - Networking Services]

Hello,

FYI, I found the issue that was in closeDefinition.yaml where the dhcp range I configured was bigger than the number of devices

 

was from 192.168.117.71 to 192.168.117.83

 

whereas only 4 devices are currently involved.

 

The DHCP server had provided (I do not know why) IP addresses for leaves that were not following the first and second provided to spine:

 

spine-01: 192.168.117.71/24

spine-02: 192.168.117.72/24

leaf-01: 192.168.117.77/24

leaf-02: 192.168.117.78/24

 

I have modified the DHCP range to : from 192.168.117.71 to 192.168.117.74

 

The DHCP server assigned .73 to leaf-01 and .74 to leaf-02 and when the ztp second stage happened, openclos did teh mapping between the source IP address of snmp traps received and teh IP addresses expected for leaves (.73 and .74).

 

This should mean that there is no automatic link between IP addresses assigned by DHCP server and IP addresses expected to be source IP of snmp traps received...

 

Anyway the second ztp stage has now completed for all 4 devices, E-BGP is working between spine and leaf devices.

 

I guess the next step should be the overlay configuration setup.