Open Issues

Cosd generates a core file only with the show command consistently while the ifd scheduler map bindings do not get updated correctly or in a Transition state during the booting time, and initiates the show interface <ifd-name> name command. The cosd application recovers after the crash and does not affect any other programming. As a workaround, wait until all the interfaces are in the Up state and complete the interface binding configurations commit before triggering the show interface <if-name> extensive command. PR1604169

When a aggregated Ethernet link is brought down, a transient error message, [Error] Nexthop: EalNhHandler: failed to add Nh: xxxx, type: composite, as pil add failed might be seen. There is no functional impact due to these errors. PR1570710

The PTP FPGA is reset during BIOS boot. During Linux boot, the PTP FPGA is taken out of reset and the pcie-tree is reenumerated. Hence you see the link going up or coming down during this sequence. PR1572061

A vulnerability in the handling of exceptional conditions in Juniper Networks Junos OS Evolved allows an attacker to send specially crafted packets to the device, causing the Advanced Forwarding Toolkit manager (evo-aftmand-bt or evo-aftmand-zx) process to crash and restart, impacting all traffic going through the FPC, resulting in a Denial of Service (DoS). PR1572969

Changing decapsulation of only the tunnel destination address configuration after tunnel is up might not work and end up using the previously configured tunnel destination address for the decapsulation. After the system enters this state, any further configuration changes to tunnel configuration cannot be handled. PR1575724

IPv6 based PKID enrollment operations are not supported over revenue ports on all Junos OS Evolved platforms. PR1584378

When all the Packet Forwarding Engines are in the Offline state and being in the Online state with multicast route being active during this cycle, the multicast traffic gets lost permanently due to the absence of multicast route. PR1598894

Sflow ingress sampling reports incorrect OIF and NH with MPLS traffic(MPLS-IPV4/MPLS-IPV6 with ECMP and Ultimate Hop Pop(UHP)) on the egress node. PR1602448

On PTX10016 routers with a fully loaded chassis and all the FPCs being in the Online state, if you take a SIB offline or online, the fabric link training takes around 10 minutes. PR1611753

On PTX10008 routers, few input PTP & SyncE-1pps noise transfer test frequencies might fail. The following result is the expected performance :

On PTX Evolved platforms with a mac-vrf instance configured with VLAN aware service, BUM(Broadcast Unicast Multicast) flooding issue might occur on the ESI (Ethernet Segment Identifier) interfaces and packets might get dropped on the egress PE devices. PR1624677

On PTX10008 routers, the transit CCM sessions comes up but transit oopback(LB) ping or LinkTrace(LT) PDUs does not go through. PR1632255

In a scaled environment, it might take up to five minutes to gather and display the p2mp LSP correlation information. PR1629771

On PTX10008 routers, after testing the laser-off link member to bring the aggregated Ethernet interface down, the following error message gets generated:

After configuring the warm-standby option, you must wait for three minutes before Routing Engine switchover. PR1623601

With back-to-back P2MP branch joins/leaves, out-of-order error might be generated in the Packet Forwarding Engine while processing the P2MP Multicast Nexthop CHG. It is a transient issue of Packet Forwarding Engine being not able to process MCNH CHG, which should be corrected by an immediate the Multicast Nexthop CHG and Make-before-break (MBB) sequence. PR1628049

Due to limitation in CBC FPGA, interrupts do not get generated if the Routing Engine mastership daemon stalls and a hardware timer-expiry mastership switchover gets initiated. Post switchover, you must reboot the backup routing-engine before another switchover. PR1633552

System reboot or boot up with traffic could result in init time fabric link crc errors and small percentage of traffic drop. PR1635178

Traceroute in MPLS OAM on SR over IPv6 might fail in ECMP case if the Junos OS Evolved device is in topology. This is because Linux kernel in Junos OS Evolved puts an autoflowlabel on every IPv6 packet. This flow label is transparent to daemon process, which uses a null value for it and calculates the nexthop details. The Packet Forwarding Engine however takes the flow label into account and calculates the nexthop details. This difference in calculation of the nexthop details leads to a mismatch in the path the packet takes to the destination and might cause traceroute to fail. PR1618406

In some reload scenarios, the config-sync service might fail due to xinetd limiting the SSH service, leading to the Major Application config-sync fail on node Re1 alarm status. This might cause configuration difference between the primary and backup Routing Engine. As a workaround, restart the service using the restart config-sync command. PR1629952

The evo-aftmand-bt process does not fully support request node halt <node> command. The command might cause cm-errors on other FPCs as the fabric links are not gracefully shutdown. Use the request chassis fpc slot <x> offline command to shutdown an FPC instead. PR1636271/xref>

On PTX10008 routers, FPC JNP10K-LC1201 frequently generates the following error message:

The switch from 10M to 5M ( or 1M), when both output and source is configured for GPS ports ( i.e output and source configuration), the REF status does not get updated in the table. This causes incorrect status to be displayed for the Clock Status and Clock Event. When you change the signal type to 10M from other frequencies, the DPLL remains in the Holdover state. As a workaround to restore the normal behavior, deactivate and activate the chassis synchronization command. PR1631689

On PTX10008 routers, transient JSR replication error messages 113/115 gets generated when you disable or enable OSPF. No functionality impact as NSR gets enabled again quickly on all protocols after the error messages. The error messages also gets generated when rpd-agent crashes. PR1627625

The CDA application crashes as soon as FPC boots up that results in fabricPfeE object not being published. Hence, the fabricPfeE objects do not get created or deleted. SIB endpoints cleanup are dependent on the deletion of the fabricPfeE objects. Stale SIB fabric endpoints prevent fabric link to bring up in further FPC reboots. PR1624765

On PTX10008 routers, few frequencies might fail to meet the mask. The following result is the expected performance for PTP-PTP and PTP-1PPS:

Profile Freq (Hz) PTP 1PPS NoiseTransfer_0_00391_Results PASS PASS NoiseTransfer_0_00781_Results PASS PASS NoiseTransfer_0_01563_Results FAIL FAIL NoiseTransfer_0_03125_Results FAIL FAIL NoiseTransfer_0_06156_Results FAIL FAIL NoiseTransfer_0_12313_Results PASS PASS NoiseTransfer_0_24625_Results PASS PASS NoiseTransfer_0_4925_Results PASS PASS NoiseTransfer_0_985_Results PASS PASS NoiseTransfer_1_985_Results PASS PASS PR1624478

Errors do not appear until there is a composite nexthop with two labels in the nexthop. This scenario must not appear and there is no impact in behavior.

PR1621689

On scaling system, load override and commit of baseline configuration causes the rpd to spin high on CPU. As GRPC configuration gets removed, sensors must be uninstalled. But, rpd does not respond to these telemetry sensors uninstallation requests failing the sensors uninstallation. Later, when you enable GRPC back on box and request the same sensor profile (cfg .json file used with jtimon) from collector, the rpd sends packets with higher sequence numbers. (Because sensor was not removed from the rpd earlier and is considered as drops by collectors, which rely on sequence-numbers.) PR1621347

With the scale configuration and NSR switched on, the cosd (Class of service control plane) application generates core files with multiple Routing Engine switchover. Most of the time, the cosd application recovers automatically. The status of the cosd application after the Routing Engine switchover displays the following message:

PR1620758

When you use 400G or 4x100G interfaces using 400G-ZRoptics, after the data path of the optics gets powered up and Tx laser enabled, sometimes the actual transmitting power might appear very low. Inside 400G-ZR software driver, a check-and-recovery mechanism has been implemented. If the actual Tx power appears very low after powering up and enabling Tx laser, a data path powering-down and re-init sequence gets exercised. The Tx power level gets recovered back to normal. After such hardware issue appears, the actual link up time appears longer than expected.

PR1616445

On PTX10008 routers, the rpd process generates a core file at rt_evo_free_sesid_info in the backup Routing Engine.

PR1607553

Vulnerability during the handling of exceptional conditions in Juniper Networks Junos OS Evolved allows an attacker to send specially crafted packets to the device, causing the Advanced Forwarding Toolkit manager (evo-aftmand-bt or evo-aftmand-zx) process to crash and restart. This impacts all the traffic going through the FPC, resulting in a Denial of Service (DoS). For for information, please refer to https://kb.juniper.net/JSA11188. PR1572969

When an aggregated Ethernet interface link is brought down, a transient error message [Error] Nexthop: EalNhHandler: failed to add Nh: xxxx, type: composite , as pil add failed might be generated. There is no functional impact due to these errors. PR1570710

On Junos OS Evolved platforms, if a firewall filter has both SCU (Source Class Usage) or DCU (Destination Class Usage) matches and forwarding-class/loss-priority matches used on the same term in the from action, the traffic supposed to be subjected to the firewall filter might not get subjected to it. PR1595788

Others FPCs gets rebooted if you restart FPC 2. PR1625065

The show snmp mib walk jnxBoxAnatomy command lists partial output. PR1628788

You must disable the inline BFD processing and only allow the delegate processing. PR1630797

During system boot up, the following PCIe Bus Error error message indicating on pcieport 0000:00:02.2 gets generated:

At some corner cases, the MPLS family filter matches fails in the latest Junos OS Evolved Release 21.4R1 images, which was due to uninitialized variable during the filter programming. Deactivating and activating the MPLS filter resumes the functionality. PR1634644

When the Packet Filtering Engine triggers reset, there is MPLS LSP transit traffic that gets wrong encapsulation and traffic gets dropped in the down stream. PR1635130

The UDP export from FPC does not work for the /junos/system/cmerror/counters/ and /junos/system/cmerror/configuration/ sensors files. The UDP export of these sensors works on the fixed-form-factor systems like PTX10001 and PTX10003. However, the UDP export of these sensors does not work on the chassis-based (modular) systems like PTX10008, PTX10016, and PTX10004. GRPC and GNMI export of these sensors do not get affected. PR1638262

When you use a source IP address as the management interface with RPF check set to strict on the interface, the response for the ICMP ping from the peer on the management interface gets dropped by the Linux kernel. The Linux kernel expects the path to the peer to be on the WAN side. PR1498255

GRES gets triggered through the request chassis routing-engine master switch command that starts displaying the following connector driver overlay messages:

Rebooting the PTX10003 router during a broadcast storm on the management port might cause a fault on the router. PR1423216

The local switching traffic sequence number does not reset. PR1560111

On PTX10003 routers, SNMP mib walk on jnxLEDTable fails. PR1620398

From Junos OS Evolved 20.4 Release, there are two different grpc python files for each JAPI file. The names of the files are *pb2_grpc.py and *pb2.py. The stub creation functions are present in the *pb2_grpc.py file. PR1580789

In Junos OS Evolved platforms, the LDP session does the authentication key-chain configuration based on the session remote-id on initiator stop from the session establishment even though you configure the authentication key-chain of the responder for the responder remote-id. PR1592431

The mgd process might crash when you configure an invalid value though JSON and XML load for the identityref type leafs or leaf-lists while configuring Openconfig or any other third-party YANG. This issue occurs with JSON and XML loads. PR1615773

This scenario occurs when you enable the IS-IS authentication key-chain that have multiple keys between the routers. The IS-IS adjacency is up since both routers have the same key active. When you manually change the system time in such a way that the routers have different keys active in the key chain, the IS-IS adjacency must go down, which does not happen. PR1572441

For the list definition list bar { key foo; leaf foo { type uint8; } leaf baz { type string; } }, Junos cannot parse the { "bar": [ { "baz": "zig", "foo": 123 }, { "baz": "zag", "foo": 0 } ] } configuration payload where leaf foo (key as per Yang model) does not appear as a first element. Junos understands only { "bar": [ { "foo": 123, "baz": "zig" }, { "foo": 0, "baz": "zag" } ] } kind of configuration payload as shown below where leaf foo (key as per Yang model) appears as a first element. As per Junos, the list keys are always required (must always) to precede any other siblings within a list entry and appear in the order specified by the data model. PR1616216

In a rare event after configuring and deleting the DDL commands, the configuration object correctly gets out of scope. However, the configuration object does not gets released due to a shared reference set chassis aggregated-devices set interface ae1 aggregated-ether-options. There is no functional impact and the situation can be cleared by restarting configuration daemon from CLI--> configd > restart configd. PR1617667