Citrix NetScaler- SDX/VPX Discovery fails with empty SNMP tables & missing identification attributes

AbdulBadi
Tera Contributor

Hi everyone,

I am working on discovering Citrix NetScaler SDX and VPX appliances.

SDX Discovery through SSH is now working after the required access was provided. The remaining issue is with VPX Discovery through SNMPv3.

The VPX responds to SNMP, returns the NetScaler system OID, and triggers the out-of-box NetScaler Load Balancer pattern. However, several pattern steps return an empty or null response, including the IPv4 address table, IPv6 address table, physical inventory table, and a Citrix DNS table.

The pattern later reaches IRE, but the VPX CI is rejected because the payload does not contain enough matching attributes for the Load Balancer identification rule. The payload contains a generic name and device description, but important values such as the serial number and management IP are missing.

The SNMPv3 account is configured for read-only access. I am now trying to determine whether the SNMP view is missing access to the standard MIB and Citrix enterprise MIB branches, or whether there is another NetScaler-side configuration required.

Has anyone encountered this behavior with NetScaler VPX Discovery?

I would especially appreciate guidance on:

  • Which SNMPv3 views or OID branches must be readable for the out-of-box pattern & how?
  • Whether both the standard MIB branch and Citrix enterprise branch can be included in one read view or how it can be done?
  • How to verify the returned values using SNMP walk
  • Whether any additional NetScaler setting is required for the serial number and management IP tables?

I have confirmed that the pattern triggers and that SNMP authentication succeeds. The current blocker is incomplete SNMP table output and the resulting IRE identification failure.

Thank you.

2 REPLIES 2

kmohammed
Tera Guru

Hi, your diagnosis is pointing the right way: classification works because the System branch is readable (sysObjectID / sysDescr under 1.3.6.1.2.1.1), while the later steps come back empty because the SNMPv3 view bound to your read-only user does not include the branches those steps walk. This is a VACM view problem on the NetScaler, not a pattern or credential problem - authentication succeeding and tables returning empty (rather than an auth error) is the classic signature.

 

Branches the OOB NetScaler Load Balancer pattern needs readable:

- MIB-II System: 1.3.6.1.2.1.1 (already working)

- MIB-II Interfaces / IP: the interface table 1.3.6.1.2.1.2.2 and the IP address table 1.3.6.1.2.1.4.20 - these feed the IPv4/IPv6 address steps and the management IP the payload is missing.

- ENTITY-MIB physical table: 1.3.6.1.2.1.47.1.1.1 - the serial number column here (entPhysicalSerialNum, .47.1.1.1.1.11) is where the physical inventory step gets its value.

- Citrix enterprise branch: 1.3.6.1.4.1.5951 - the NetScaler private MIB (system/hardware and the Citrix-specific tables the pattern queries, including its DNS/table data).

 

One view or two? One view is fine and is the normal setup - a VACM view is a list of included subtrees, not a single branch. On the NetScaler, extend the view assigned to your SNMPv3 group/user so it includes both 1.3.6.1.2.1 (or at least the specific standard tables above) and 1.3.6.1.4.1.5951 as included subtrees. Read-only is sufficient; nothing here needs write access, and no extra NetScaler feature/setting is required for the serial number or management IP beyond those branches being readable.

 

Verify with SNMP walk before re-running Discovery, using the same MID Server host and exactly the same v3 user/auth/priv as the credential: walk 1.3.6.1.2.1.4.20, then 1.3.6.1.2.1.47.1.1.1.1.11, then 1.3.6.1.4.1.5951 against the VPX NSIP. Three outcomes tell you where you stand: 'noAccess / not in view' style responses confirm the view is still the blocker; real values in the walk but empty steps in the pattern mean you should then compare against the OID each pattern step shows in the ECC queue probe/pattern log; values in the walk and the pattern will then normally populate the payload and satisfy the identification rule.

 

One VPX-specific caveat on the serial number: a VPX is virtual, so if the ENTITY-MIB serial column is still legitimately blank after the branch is readable (some virtual appliances do not present a physical serial), do not try to force one - in that case make sure the management IP from the IP address table is returned, because name + IP from the payload gives IRE something real to match on, whereas an appliance with neither a serial nor an IP in the payload will keep being rejected however the view is set.

Hi @kmohammed 

 

Thanks for the detailed explanation.

 

My original post is a week old, so here is where things stand now.

 

Now working. SNMP Discovery retrieves the serial number, management IP, interfaces, VIPs, pools and pool members:

  • Serial comes from the NetScaler enterprise OID 1.3.6.1.4.1.5951.4.1.1.26's sibling 1.3.6.1.4.1.5951.4.1.1.14.
  • IPs come from 1.3.6.1.4.1.5951.4.1.1.26.1.
  • The standard interface table also returns data.

Still empty, in both the pattern and an external SNMPv3 walk with the same credentials:

  • IPv4: 1.3.6.1.2.1.4.20.1
  • IPv6: 1.3.6.1.2.1.55.1.8.1
  • Physical inventory: 1.3.6.1.2.1.47.1.1.1.1
  • Citrix domain table: 1.3.6.1.4.1.5951.4.1.1.51.3.1.1

The configured view from the device Side includes 1.3.6.1. I believe this should cover all in general.

 

As you suggested, we're verifying the actual user → group → view assignment and any exclusions.

 

If those are correct, are these tables expected to be implemented and populated on NetScaler VPX 13.1 (build 64.32)?

 

The NetScaler docs I found list only a subset of standard MIB-II groups. Have you seen a specific configuration or known agent issue that would explain selective empty responses like this?

 

Any build-specific documentation would help us decide whether a device change is needed, or whether the pattern should treat these tables as unavailable.

 

Thanks,
Abdul Mohammed