Open Issues

Default FC names (Factory defaults) can't be redefined in Junos-EVO. Different/Custom FC names can be used as an alternative. This has no bearing on the functionality. PR1827230

When DHCP trace options are enabled, there is a possibility that jdhcpd could core. In general, traceoptions should be enabled only for debugging. They should be disabled once debuffing is done.PR1771121

On all Junos and Junos evolved platforms with telemetry enabled, if the streaming server and export profile for reporting-rate are not properly configured in the analytics settings, rebooting the FPC would prevent any of the interfaces from coming up. PR1779722

On Junos Evolved PTX platforms, the log message "CFMMAN: Parse Error: CFM CCM pkt TLV parsing error" may appear when an FPC or PFE is restarted and comes online. This message indicates that a CFM Continuity Check Message (CCM) was truncated, leading to an error while parsing the TLV (Type-Length-Value) fields in the packet. The error is transient and typically affects only a small number (usually 1 or 2) of CCM packets received immediately after the FPC or PFE becomes operational. It has no functional impact on CFM sessions or overall network operations. CFM continues to function normally, and no persistent degradation is observed. This issue does not consistently affect any specific CFM session and occurs only during the brief initialisation period of the hardware component. PR1810549

On Junos OS Evolved based platforms, while using ping, traceroute, or other utility that requires host name resolution, an error is raised indicating hostname resolution has failed.PR1822994

RPC and CLI validation failed for cli show ethernet-switching table. PR1825059

On all Junos OS Evolved platforms when recursively resolved routes are changed or deleted, route churn potentially leads to the rpd process crash. PR1830588

On Junos OS Evolved PTX10003 platform the command indirect-next-hop-change-acknowledgements is required in the junos-default configuration. If the said command is missing, it will result in packet loss. PR1836337

On Junos OS Evolved PTX platforms, when subinterface is disabled using command set interfaces <name> unit disable, packets are still getting forwarded since IFL state remains up in ASIC. As a workaround, instead of disabling the subinterface, it should be deactivated using command deactivate interfaces <name> unit <> so that the subinterface is removed and packet flow will be stopped. PR1843667

When multiple SIBs are offlined in succession, at times it can result in continuous traffic drops. PR1849563

On PTX EVPN MPLS-VXLAN Stitching DC-GW, vxlan traffic might select NH from inet3 table, which will cause traffic loss. In order to avoid this, increase LDP preference to use inet.0 table to resolve the NH for vxlan traffic. PR1853669

On all Junos OS Evolved platforms, enabling the BGP RIB-Sharding causes the routing protocol daemon (rpd) leading to spike and remain at 100% CPU usage due to a background task (such as the Route Target/User Interface (RT/UI) delete job ) entering into the continuous processing cycle and looping behaviour. PR1854481

DNS resolution for traceroute does not work for a router using the mgmt_vrf with 23.4R2-EVO. PR1858650

When two or more SIBs are offlined/restarted in quick succession, transient host loopback wedge major alarms can be seen. PR1860468

SSH issue from the Routing Engine to the FPC causes SIB HA timeout error upon Routing Engine switchover. All the SIBs reset as a result of that. PR1860970

An LSI IFL remains in RPD even after being deleted by the interface manager daemon. It is visible in show interface routing but not in show interfaces, indicating that RPD still holds the IFL despite its removal elsewhere. rpd-agent does not send a delete message to RPD due to a reference count issue. Another daemon?likely l2ald?still holds a reference to the IFL. rpd-agent only sends the delete once all references are cleared, which doesn't happen in this case. The fix is to send a delete pending message from rpd-agent to RPD. RPD will treat this as a delete and remove the IFL, ensuring consistency across the system. PR1866522

On all Junos OS Evolved platforms, a failure in internal process communication causes the system to stop responding to the show system applications app ndp detail CLI command or prevent applications from start/stop/restart. This issue occurs when systemd becomes unresponsive to requests initiated by sysman. There is no impact to traffic forwarding. PR1866988

Configure family inet6 for interfaces which are part of v6 over v4 hierarchies like follows: v6 Route (RTT_USER) -> Indirect(44168)-> Unilist(44160) -> v4 Unicast(44152). PR1870118

On Junos OS Evolved PTX10001-36MR platform with PTP (Precision Time Protocol) configured, a synchronization issue occurs between the PTP FPGA (Field Programmable Gate Array) and the MAC (Media Access Control)/PHY (Physical Layer). This results in timing errors on the local device, which might also propagate to downstream devices. The issue typically arises after restarting the evo-aftmand-bt or evo-cda-bt processes, which impacts the socket connection between these processes and the timingd service. As a result, timingd is no longer able to communicate with these components as expected. PR1878029

MACsec traffic can be impacted due to incorrect port mapping. When MACsec logical interface is configured with a custom EAPOL ethertype profile on dual VLAN tagged interfaces, traffic fails in one direction while the MACsec session appears up. This issue is observed during WAN MACsec use cases on PTX10002-36QDD and PTX10003-80C platforms. The issue does not occur when MACsec logical interface is disabled. PR1879375

During link detraining sometimes cru_err_pg_psc interrupt may be present for a very brief time in SIB. It is not a problem if its seen during offline. But should not occur during stable state. PR1880275

On Junos Evolved PTX platforms, when configuring a firewall policy, specifying family any, and matching on inner IP payload (IPv4 or IPv6), matching does not work on Multiprotocol Label Switching (MPLS) packets. PR1882315

In Junos OS Evolved, ARP resolution requests are throttled on FPC per logical interface level that is if resolution request on a logical interface expires then a throttle timer is started on that logical interface and no other resolution request could be generated on that logical interface when throttle timer is running. In this issue RE netstack is sending packets to FPC with hint to resolution request for already resolved IP address and triggering throttle timer. So resolution request for second IP address on same logical interface could not be generated for some time which is triggering phone-home application 10 seconds time out. PR1883158

This issue is likely to occur when the interfaces are activated/deactivated/created/deleted. This could be due to a) Bulk change in interface configuration or b) Admin-initiated application restart via command request system application node re0 restart app <application> c) System triggered automatic application restart due to internal events. Note that this issue occurs due to very rare combination of events and would not be seen in most use cases. There is no service impact when the issue occurs, and router operations are unaffected. PR1885455

This is a transient log which sometimes is seen when LSI is recreated. This has no functional impact. It's generated because of additional dependency of LSI with BD. System takes care of cleaning the token for which this error is generated. This can be confirmed through VTY command > show sandbox token. <> PR1834443

In case of PTX, continuous MAC move events of same MAC between two or more interfaces will take a worst case maximum of 2 to 3 secs, as the TOE block in PTX waits for an ACK from host before it can move the same MAC again to a new interface. The delay is due to the fact that the host ACKs are processed by the TOE block in a two second timer thread. Also, in the BX chipset, there is an additional FIB cache cleanup logic delegated to host, which can add bit more delay in clearing the FIB cache entries in DLU block, to allow new events for the same MAC. Until then the TOE thread will not allow the same MAC to be moved. Due to the above said limitation, the control plane in PTX cannot record more than one MAC move event within the same second. So, to detect and record MAC move events, the below configuration is suggested. set protocols l2-learning global-mac-move threshold-time 10 set routing-instances <ri-name> switch-options mac-move-limit 1 set routing-instances <ri-name> switch-options mac-move-limit packet-action logPR1839324

This hdr_crc_err in FI (Fabric Input block) is due to CRC error corruption in header part of the cell that gone past beyond the FEC. This will result in cell drop. When FPC (LC1301) is starting up user might see dp_X_fi_X_intr_pcs0_hdr_crc_err in inactive system errors. This interrupt happens during the link training phase before the switching core logic input is enabled. So, this cell with hdr_crc_err is dropped and not processed any further. This is expected as the preparation of the link is done earlier, however this will not impact the system as FI block will not be enabled, when this interrupt is raised, during training. This note is applicable for LC1301 and SF5 SIB hardware combination.PR1873313

If FPC is yanked out ungracefully out_of_range_spry_indx interrupts may continue to come. This will stop once FPC is online back again.PR1875117

[fabric] [fabrictag] : Junos OS Evolved PTX10008: some ccl is not fixed by AH, when GRES is initiated while AH is in progress.PR1843391

With pre-gnmi telemetry (telemetrySubscribe) streaming, when the subscription is made with need_eos flag set to false, where initial sync is not requested, there is a possibility that counters with zeros are also will be sent atmost once. If collector is not interested in those zeros counter can ignore those data., as its streamed only once. This is not an issue if collector is requesting with need_eos set to true in the request as zeros counters are already being sent atleast once in this scenario. PR1843391

On PTX10003 systems, it is not allowed to configure ZR optics (400G) through CLI. PR1851078

On a PTX10016, and possibly PTX10008, the router shows fabric link OAM fault causing the system to not function properly. PR1900242

Multiple policies under same routing instance when applied on interfaces were not working correctly as this was not supported earlier. Now support has been added .PR1873569

On all Junos OS Evolved platforms, when two policers with the same initial name(First 15 characters of rate estimator name) and term names are configured, since the rate estimator name (combination of filter term name and policer name) which gets truncated to 15 characters, the two policers start using the same rate estimator and it can lead to possible early drops by the policer, based on the current traffic rate of the two policers. PR1896496

The RPD process crashes on all Junos and Junos OS Evolved platforms in a highly scaled scenario of more than 2000 BGP peers if the BGP sessions are un-configured/brought down abruptly. This leads to loss of routing information and will lead to loss of protocol traffic. PR1756603

On all Junos and Junos Evolved platforms with dual Routing Engines (REs), Protocol Independent Multicast (PIM) neighborship is not be maintained on the backup Routing Engine after a ppmd-agent restart. This can lead to loss of PIM neighbor state on the backup RE. PR1880262

When a device is rebooted with PAA installed in 25.2R1 , PAA installation on reboot may fail due to "Not found default vrf" error. This can be resolved by deactivating and activating or freshly installing the PAA configuration. PR1886928