How to discover all peer nodes completely?

[victor]

Hi!

It could be related to LLDPv2 MIB support that we have added in 4.3. Could you please check if device responds to OID .1.3.111.2.802.1.1.13.1.3.2.0? If yes, could you provide result of SNMP walk on .1.3.111.2.802.1.1.13.1.4.1.1 and .1.0.8802.1.1.2.1.4.1.1?

Best regards,
Victor

[justrest]

Thank you very much for your reply! 

I'm on a business trip and can't import the list, but I had a colleague try and the SNMP walk results for .1.3.111.2.802.1.1.13.1.4.1.1 and .1.0.8802.1.1.1.2.1.4.1.1 seems containing  all the neighbor information, just some information that shows up as garbled. The current codepage setting of snmp is CP936 (Chinese), could this problem be related to the codepage setting?

[justrest]

@victor said in How to discover all peer nodes completely?:

Hi!
It could be related to LLDPv2 MIB support that we have added in 4.3. Could you please check if device responds to OID .1.3.111.2.802.1.1.13.1.3.2.0? If yes, could you provide result of SNMP walk on .1.3.111.2.802.1.1.13.1.4.1.1 and .1.0.8802.1.1.2.1.4.1.1?
Best regards,
Victor

Thank you very much for developing such an excellent network management software, could you please help to optimize the new version of LLDP protocol support, thank you very much!

[victor]

Yes, it actually could be related to codepage. Depending on entry type, NetXMS server may require to search for peers using texts (interface descriptions for example), and if it is received incorrectly search will fail. To be able to diagnose it I likely will need exact walk outputs though. We can move discussion to private messages or Telegram if you don't like to share such data publicly.

Best regards,
Victor

[Egert143]

I don't want to hijack a thread, but for example, Cisco C9200L-24P-4X switch peer discovery doesn't work 100%. Only other Cisco switches are discovered, but no other device directly connected to the switch. It broke with a newer version of NetXms, but I don't remember which version exactly.

[justrest]

@victor said in How to discover all peer nodes completely?:

Yes, it actually could be related to codepage. Depending on entry type, NetXMS server may require to search for peers using texts (interface descriptions for example), and if it is received incorrectly search will fail. To be able to diagnose it I likely will need exact walk outputs though. We can move discussion to private messages or Telegram if you don't like to share such data publicly.
Best regards,
Victor

No, actually none of this information is considered private, it just seems like too much. Thank you very much for your reply, I've uploaded the relevant attachments, so please take the trouble to guide me, thank you very much!

[justrest]

@victor said in How to discover all peer nodes completely?:

Yes, it actually could be related to codepage. Depending on entry type, NetXMS server may require to search for peers using texts (interface descriptions for example), and if it is received incorrectly search will fail. To be able to diagnose it I likely will need exact walk outputs though. We can move discussion to private messages or Telegram if you don't like to share such data publicly.
Best regards,
Victor

Currently the server configuration file netxmsd.conf as well as snmp.codepage and syslog.codepage are CP936, except that previously under version 4.2.461, the same configuration did not have this problem.

[victor]

Hi,

I've found a bug introduced in 4.3 which may cause incorrect decoding of strings received via SNMP (SNMP codepage was ignored in some cases). Fix will be included into upcoming patch release 4.3.6.

Best regards,
Victor

[justrest]

@victor said in How to discover all peer nodes completely?:

Hi,
I've found a bug introduced in 4.3 which may cause incorrect decoding of strings received via SNMP (SNMP codepage was ignored in some cases). Fix will be included into upcoming patch release 4.3.6.
Best regards,
Victor

thanks very much！！

[justrest]

@victor said in How to discover all peer nodes completely?:

Hi,
I've found a bug introduced in 4.3 which may cause incorrect decoding of strings received via SNMP (SNMP codepage was ignored in some cases). Fix will be included into upcoming patch release 4.3.6.
Best regards,
Victor

I'm very excited to try the just released version 4.3.6, but the network device still can't seem to discover all neighboring devices ...... I uploaded the latest walk result of SNMP, please help me to see it, thank you very much!

[victor]

Could you also provide server log file with debug level 6 for the duration of topology poll on that switch (i.e. set debug level to 6, go to Poll -> Topology on the switch node, set debug back)?

[victor]

Hi,

I was able to reproduce and fix this issue (yet another bug in LLDPv2 MIB handling). Fix will be included into next patch release (we likely will publish it this week).

Best regards,
Victor

[justrest]

Thanks very much！

[justrest]

@justrest said in How to discover all peer nodes completely?:

@victor said in How to discover all peer nodes completely?:

Hi,
I've found a bug introduced in 4.3 which may cause incorrect decoding of strings received via SNMP (SNMP codepage was ignored in some cases). Fix will be included into upcoming patch release 4.3.6.
Best regards,
Victor

I'm very excited to try the just released version 4.3.6, but the network device still can't seem to discover all neighboring devices ...... I uploaded the latest walk result of SNMP, please help me to see it, thank you very much!

Thank you so much for your great help! I tried the latest version 4.3.7 today and the problem with lldp v2 still seems to exist, please help me with the attached logs, much appreciated! Or is it possible to force the use of lldp v1 version? Looking forward to your reply!

[justrest]

@victor said in How to discover all peer nodes completely?:

Hi,
I was able to reproduce and fix this issue (yet another bug in LLDPv2 MIB handling). Fix will be included into next patch release (we likely will publish it this week).
Best regards,
Victor

Since under 4.2.* version, all peer nodes can be discovered normally, and the topology is shown completely (more than 30 nodes), but after 4.3 version can only discover part of the nodes (about 7 nodes), and as the time becomes longer, only two nodes are left at last,  is  it possible to turn off the LLDP V2, and revert back to the recognition mechanism of 4.2.* version, or there is any way to solve the problem. Thank you very much!

[victor]

I have made some changes in reading LLDP information in upcoming release 4.4.1. FIrst of all there were few bugs left in reading LLDP-V2 MIB. Also, now server will use both LLDP-MIB and LLDP-V2-MIB if supported by device, and LLDP-MIB takes precedence. This should solve discovery issues.

Best regards,
Victor

[justrest]

@victor said in How to discover all peer nodes completely?:

I have made some changes in reading LLDP information in upcoming release 4.4.1. FIrst of all there were few bugs left in reading LLDP-V2 MIB. Also, now server will use both LLDP-MIB and LLDP-V2-MIB if supported by device, and LLDP-MIB takes precedence. This should solve discovery issues.

Best regards,
Victor

Thank you very much！

[justrest]

@victor said in How to discover all peer nodes completely?:

I have made some changes in reading LLDP information in upcoming release 4.4.1. FIrst of all there were few bugs left in reading LLDP-V2 MIB. Also, now server will use both LLDP-MIB and LLDP-V2-MIB if supported by device, and LLDP-MIB takes precedence. This should solve discovery issues.

Best regards,
Victor

We've updated to version 4.4.1 and it looks like H3C's Neighborhood Discovery feature should be working fine, fantastic, thanks so much!