Skip to content

Troubleshooting Discovery

Documentation on most commonly seen discovery issues and how to resolve them.

Settings

The below topics will be describing the Global Settings however users can make adjustments to specific Snapshots. Making changes to the Global Settings and then testing them in a previous snapshot may not apply your changes. Here are two methods for testing settings.

  • Make your changes in the Snapshot specific Settings and then Add Devices (or Rediscover timed out devices). If these changes fixed your issue, then apply them globally.

  • Make your changes globally and then run a new Snapshot. You could also filter the allowed discovery scope in the Snapshot specific Settings to a subset of /32 addresses to speed up discovery. In the example below I am telling discovery only to test 1 device.

Discovery Seed

These are self-defined IP Addresses that IP Fabric will try to connect to and discover. If a device is successfully discovered it will check the settings in the Discovery Tasks Settings to see what technologies to use to discover neighbors.

If you have remote locations possibly through a WAN that is not owned by you, this would be the location to put the IP Addresses of some of your devices so IP Fabric can discover and crawl that site.

Important

Adding new devices here will only discover them if Limit discovery to already discovered devices is not enabled in the Discovery Tasks Settings.

Management → Discovery History

The Discovery History table is also used by IP Fabric as the seed addresses. Once a device is discovered in a snapshot it will be placed in this table for use in other snapshots.

  • The username last successfully used for a device will be recorded here and be the first one tried in a future snapshot. If unsuccessful it will check the Settings → Authentication list and go back through the decision tree. This may cause the next snapshot to take longer due to authentication failures but once this table is updated with the new username the discovery time will return to normal.
  • If you have removed devices from your network IP Fabric will still try to connect to them because the IP addresses are defined as seeds in this Discovery History table. To speed up discovery you could delete stale records. You can also filter on Last discovery time and delete records older than X months.

Settings → Advanced

Discovery → IP Scope

IP Scope tells IP Fabric which networks to include or exclude in the discovery process. The default is set to include everything (0.0.0.0/0). When IP Fabric discovers a neighbor in this network it will try to log in and discover it. If the IP address is not in the include or is in the exclude list then it will not try to connect and discover the device.

The exclude list is a great way to exclude networks managed by a different department or perhaps vendors or devices that IPF does not currently support. For instance a customer had discovery hanging due to it trying to log into a terminal server which IPF could not understand; once added to the exclude list the discovery completed successfully.

Important

Do not confuse the Include List with the Seed list. Include will only try to discover an IP if it is found through the Discovery task or the IP in the seeds. Also note IP Fabric does not do any ICMP pings to find hosts so having a /16 here will not send massive amounts of pings.

Discovery Tasks Settings

Discovery Tasks Settings is where the magic happens for automated discovery of new devices on your network.

  • If you are not finding new devices in your snapshot ensure Limit discovery to already discovered devices is not enabled. If this is enabled then only devices in the Discovery History table will be added to the snapshot and no new devices (even if those devices are manually added to the Settings → Discovery Seed).

    If you want to limit to only discovered devices but want to add new devices this is still possible by going to Discovery Snapshot and manually adding devices into a snapshot which it will then be added to the Discovery History table and picked up in future snapshots.

  • xDP (neighbors) signifies using CDP or LLDP information to discover devices in your network.
  • ARP uses the ARP and MAC Address OUI information to find devices. If the OUI is set for Enabled for discovery in the Settings → OUI table then IP Fabric will attempt to connect and discover the device.
  • Routing Table will try to connect to next-hop devices.
  • Trace signifies using traceroute to RFC1918 addresses to help discover your internal network. This is helpful for the VM as IPF might not be able to log into the default gateway (vRouter) but using traceroute it can find some other physical devices in the network to use as starting points.

SSH/Telnet

Please see SSH/TELNET for explanation of these settings.

Vendors API

Please see Vendors API as each vendor has different requirements.

Settings → Authentication

Authentication is where you define the username and password IP Fabric uses to connect a physical device (devices discovered through the API are managed through Advanced → Vendors API). Ensure that you have a username configured for all scopes of the network you wish to discover or set to the default of 0.0.0.0/0.

Tip

If you are having issues with Configuration Backup not pulling data ensure that you have Use for configuration management set on the proper usernames.

Further information can be located at Authentication.

Troubleshooting Missing Devices

There can been a numerous amount of reason why IP Fabric did not discover a device from Day 0 or incorrectly configured devices, AAA outage, or insufficient authorization privileges to name a few. First we will take a look at some tables to help you find these and then show how to use the Connectivity Report to debug them.

Vendor Specific Troubleshooting

You may notice that some devices of a certain vendor are not being discovered. Please read through the Known Issues to ensure a bug has not been already raised or a fix suggested. For instance some Firewall vendors require elevated or admin profiles in order to pull all data.

Technology Tables

CDP/LLDP → Unmanaged neighbors

This is the best table to start as not only do you see the local host and interface names but also the remote hostname, interface name, and remote IP. Since devices can connect to multiple other devices it is possible of duplicate remote hostnames in this list.

You are able to export this view to CSV and then remove duplicate hostnames to get a unique list of devices. For instance in this demo I see a total of 59 unmanaged neighbors but only 40 unique remote hostnames.

Interfaces → Connectivity matrix → Unmanaged Neighbors Detail

This table shows you unmanaged neighbors based many protocols and has a built-in default Intent Check for you.

  • RED: Provides a list of unmanaged CDP/LLDP neighbors (cdp|lldp|mndp). This does not provide you with the remote hostname which is why it is recommended to look at this table second.
  • AMBER: Is a list of unmanaged Interior Gateway Protocols (eigrp|ospf|rip|isis|pim|vxlan|cef). It is specified as such because if you have a neighbor adjacency with an iGP it is assumed it is under the control of your network.

    Tip

    Check this table for help locating rogue devices in your network.

  • BLUE: Is the default rule if the protocol does not match any other regex.
  • GREEN: Is defined for the Exterior Gateway Protocol BGP.

Auditing using an External Network Management System

Perhaps you would like to audit IP Fabric (or vice versa) using an NMS. This can be done as all the tables can return the data in CSV format via the UI or JSON using the API. This would perhaps help you find other missing devices IP Fabric could not discover.

Tip

Don’t forget to audit your NMS or alarming systems inventory from what IP Fabric discovered on the network. Many customers have found devices that were not being monitored.

Note on Management IP

In IPF v4.3.5 or below the UI in many places has a column of Management IP; in v4.4 this has been renamed to Login IP to avoid confusion and to match the API column name of loginIp.

Since IP Fabric is an automated discovery process it will try to log into any IP address found during the discovery tasks processes. This value could be a Loopback, management, physical, or virtual interface. Auditing your external NMS will need extra care as in many cases the IP’s will not match.

One solution is to export your NMS inventory to Excel and the Technology → Addressing → Managed IP table to CSV and then perform vlookups between the two datasets. Since the Managed IP contains all the IP addresses found on devices the NMS data should find a match.

Another option is to use the device’s Serial Number. Serial numbers are not always unique and there is a chance of overlap so please take this into account.

Note on Serial Numbers

IP Fabric records two Serial Numbers for the device. The column name Serial Number is the actual hardware serial number of the device which is why it is labeled snHw in the API.

The column name Unique serial number is an IP Fabric unique identifier and primary key for the device. This API column name is sn so it is easy to confuse the two if you were not aware. The main time there are differences between the two is seen in firewalls with virtual contexts, vdoms, etc.

Connectivity Report

The Connectivity Report is a per Snapshot report stating the successes or errors of IP addresses that IP Fabric tried to discover. Once you have a list of IP Addresses from the unmanaged neighbor tables above or perhaps an external Network Management System this is where you would start your troubleshooting process.

The most common reason why a new installation of IP Fabric will not connect is due to the connect ECONNREFUSED XX.XX.XX.XX:22 error which signifies the traffic is being blocked by an ACL or Firewall. This is why it is recommend having IP Fabric in your management subnet which is already allowed for remote access to devices so you do not have to re-configure all your ACLs. Another option is to use a jumphost and documentation can be found at Jumphost settings.

Another common error is due to Authentication or Authorization errors. Please ensure that the username configured has the correct permissions. There is also a list of CLI commands used which can help ensure authorization is correct.

If you find other errors you can refer to this document Connectivity report - Type of Error and the CLI output of the device is usually helpful as well.

If you need further assistance please feel free to reach out to your Systems Engineer or open a ticket following the Technical Support instructions.

Finally another option for testing devices is to log into the IP Fabric CLI using the osadmin account and manually trying SSH’ing into your devices preferably using the username and password the system was configured to use. If you cannot open a connection this is due to an external reason.

Further information can be found at No Devices Discovered.

Troubleshooting Configuration Management

Configuration Management is a separate process from discovery. First a device must be discover in a Snapshot and placed in the Management → Discovery History table. Once in this table then next time IP Fabric is scheduled to pull configs it should be populated in the list.

Example

  • Settings → Advanced → Snapshots is set to periodically create snapshots every day at 6AM
  • Settings → Advanced → Configuration Management is set to periodically create snapshots every day at 11PM
  • New device will be found at 6AM but since the config mgmt is scheduled for 11PM you will need to wait until the next day to ensure it was pulled.

Other troubleshooting tips:

  • Please ensure that in Settings → Authentication you have enabled the correct usernames to Use for configuration management, if no usernames have this enabled then configs will not be backed up.