Table 3 lists known issues in NorthStar Controller Release 5.1.0. If an identifier is reported, it is the assigned identifier in the GNATS problem report tracking system.
Table 3: Known Issues in NorthStar Controller Release 5.1.0.
Junos OS: A user can configure a template in the router and map that template to an external controller. The router inherits the required configuration from the template and then provisions the external controller-initiated LSP. Unbinding the template from the external controller or changing template configuration can trigger deletion of the PCE-initiated LSPs (only LSPs which are using that particular template). Later, the LSPs are re-provisioned by the external controller.
Before performing a fresh install of NorthStar Release 5.0.0, you must use the ./uninstall_all.sh script to uninstall any older versions of NorthStar on the device.
PRPD does not remove prefixes (prefixes that have mapping) that were withdrawn by PCCs.
P2MP: During transient make-before-break, the PCE server might erroneously send a delete event message to the PCS, potentially causing a branch to go from “live” to “unknown” state. The PCS then detects the unknown state and tries to reprovision. The PCE server accepts the order and forwards it to the PCC, but the PCC returns an error saying the LSP already exists. The only workaround is to deactivate and activate the PCEP protocol.
See “PCS Out of Sync with Toposerver” in the NorthStar Controller Troubleshooting Guide for instructions to recover when the PCS is out of sync with Toposerver.
P2MP: LSPs are being deleted when they should be modified.
NorthStar cRPD does not forward adjacency-SID data to the topology server. As a consequence, cRPD cannot be used for applications that use segment routing (including EPE steering). This is due to Junos OS behavior.
NorthStar web UI Planner: Network spec files will be overwritten if an existing network name exists when using Save or Save As. A warning dialog appears if an existing name is found using Save As.
NorthStar web UI Planner: If a user leaves the browser open without activity for a period of time, the session may get disconnected and not be able to fully resume later. As a result, some UI operations may not response properly. The workaround is to use the close network menu option and reopen the network.
The routing method is CSPF for P2MP groups which are configured from devices (as opposed to from within NorthStar), when the expected routing method is routeByDevice.
NETCONF-Provisioned SR Tunnels: NETCONF-provisioned SR tunnels can only depend on static (as opposed to dynamic) SIDs. Dynamic SIDs can change at any time due to various network events such as interface flaps. If SIDs change, NorthStar detects the tunnels affected and marks their paths as “down”, but still shows their tunnels as “up”. The paths for these tunnels are no longer highlighted in the topology.
For IOS-XR devices, you must run device collection before doing any LSP delegation. This applies to LSPs that were manually created using the router CLI.
NETCONF network information table update: when you delete an interface, the interface list is not updated in the Interface tab of the network information table.
PRPD does not update Top N Prefixes properly.
Normally, the web UI automatically switches to safe mode, and allows the admin user to log in when the Cassandra database is down. In this release, the automatic switch does not work. The workaround is to restart web server when you cannot log in to the web UI during a Cassandra failure.
NorthStar Planner Desktop: There is no validation on the NorthStar Planner Destop when a license upload is attempted.
The privateForwardingAdjacency (a pair of binding SID SR LSPs from A->Z and Z->A) logical link could be marked as down after the TopoServer process is restarted. The current workaround is to run a Device Collection task in the Task Scheduler, which would restore the correct status to the privateForwardingAdjacency logical link. This issue will be addressed in the next release.
Diversity group does not work as expected for individually-created LSPs assigned to the same diversity group. We recommend provisioning diverse LSPs using Applications > Provision Diverse LSP.
EPE Planner application can fail to start after a fresh install of NorthStar Controller. Sample log messages when this happens:
To recover, restart the EPE Planner process:
# supervisorctl restart epe:epeplanner
PCViewer does not update PCE-controlled LSP paths, resulting in LSP delay not being calculated and updated in the NorthStar web UI.