Open Issues

Earlier implementation of kvmclock with vDSO (virtual Dynamic Shared Object) which helps avoid the system call overhead for user space applications had problem of time drift, the latest set of changes takes care of initializing the clock after all auxiliary processors are launched so that the clock initialization is accurate. PR1691036

in a H-VPLS network with VPLS hot-standby and the routing-options forwarding-table vpls-hotstandby-convergence command enabled on spokes, if the active hub is rebooted, 20-25 seconds loss for inter-zone traffic stream is seen. This is due to hubs in other zones connected by full-mesh ldp, starting global repair before spokes starting local repair. PR1699645

Tag rnh appears to be freed somewhere in the corner case, but the relevant pat node has been missed to delete from the tag patricia tree. That makes tag rnh/(pat_node->Tnh) a dangling pointer and later on, it results in a crash while accessing invalid pointer addresses in the tag rnh/Tnh structure. PR1707053

Traceroute in MPLS OAM on SR over IPv6 may fail in ECMP case if EVO box is in topology. This is because linux kernel in EVO 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 NH details. PFE however takes the flow label into account and calculates the NH details. This difference in calculation of NH details leads to a mismatch in the path the packet takes to the destination and can cause traceroute to fail. PR1710285

On all Junos and Junos OS Evolved platforms (For QFX5100, only in Virtual Chassis-VC setup) with RSVP (Resource Reservation Protocol) LSP (Label-Switched Paths) configured in multi vendor deployment and Juniper routers is acting as a transit/ingress routers and RESV (Reservation Request) message is received with RESVCONF object from other vendors, rpd process crash will be observed. PR1723229

In some NAPT44 and NAT64 scenarios, duplicate SESSION_CLOSE Syslog occurs. PR1614358

YANG: After upgradation s/w version on DUT, yang package with lower revisions are available in upgraded s/w version. PR1693646

When the "deactivate services rpm" and "deactivate routing-options rpm-tracking" clis are applied together and then committed, some of the rpm tracked added routes are not deleted from the routing table. Issue cannot be seen using the following steps. 1. deactivate routing-options rpm-tracking 2. commit the configuration then all the rpm tracked routes will be deleted. If the RPM service needs to be deactivated, 3. deactivate services rpm 4. commit. PR1597190

If a vmhost snapshot is taken on an alternate disk and there is no further vmhost software image upgrade, the expectation is that if the current vmhost image gets corrupted, the system boots with the alternate disk so the user can recover the primary disk to restore the state. However, the host root file system and the node boots with the previous vmhost software instead of the alternate disk. PR1281554

VXLAN VNI (multicast learning) scaling on QFX5110 traffic issue is seen from VXLAN tunnel to Layer 2 interface. PR1462548

Runt, fragment and jabber counters are not incrementing on EX4300-MPs. PR1492605

VE and CE mesh groups are default mesh groups created for a given Routing instance. On vlan/bridge-domain add, flood tokens and routes are created for both VE and CE mesh-group/flood-group. Ideally, VE mesh-group doesn't require on a CE router where IGMP is enabled on CE interfaces. Trinity based CE boxes have unlimited capacity of tokens, so this would not be a major issue. PR1560588

On EX2300, EX3400,:EX4300-48MP and EX4300 , Pause frames counters does not get incremented when pause frames are sent. PR1580560

Pim Vxlan not working on TD3 chipsets enabling VxLAN flexflow after release 21.3R1. Customers Pim Vxlan or data plane VxLAN can use the version 21.3R1. PR1597276

output of show network agent command should be null, which shows statistic per component after GRES. PR1610325

For a topology with VSTP and VRRP configured and IPV6 traffic, if VSTP bridge priority is changed a couple of times (to trigger toggling of root bridge), it is possible that V6 traffic drop is seen on some of the streams. PR1629345

mspmand daemon running on MS-MPC/MS-MIC cards can occasionally crash when the service card (fpc/pic) is turned offline and then online at regular intervals when the number of service-set configured is moderately high and when extensive hardware crypto operations are being performed. Exact issue is yet to be isolated. PR1641107

Please do not enable host-path tracing when there is high volume of packets been received in the host-path. PR1645741

If the physical link status of the ethernet link between the RE and FPC goes down, there are recovery attempts to bring up the link again. Log messages indicate the recovery attempts and the success/failure status of the attempt. However an alarm is not raised when this failure occurs. PR1664592

In case Port is DOWN then Tx Laser need to enable via cli-pfe> prompt.PR1673892

There will be drop of syslog packets seen for RT_FLOW: RT_FLOW_SESSION_CREATE_USF logs until this is fixed. This will not impact the functionality. PR1678453

On QFX5100 platforms (both stand-alone and VC scenario) running Junos, occasionally during the normal operation of the device, PFE (Packet Forwarding Engine) can crash resulting in total loss of traffic. The PFE reboots itself following the crash. PR1679919

The issue here is that we see ?MQSS(0): DRD: Error: WAN reorder ID timeout error? once per PFE during bootup of FPC. This happens because during the FPC bootup some control packet from vmhost comes before the PFE init is fully complete. Because of this the EA Asic is not able to process the packet and throwing the error. The fix involves complex changes in the bootup sequence of ASICS and will result in other major issues. The original issue has no functionality impact. It is just one error per PFE seen during the FPC reload case only. At that time the traffic is not started yet and once the system is up no other impact is seen due to the Error. Hence the issue will not be fixed. Any "WAN reorder ID timeout error" during the bootup of FPC can be safely ignored. PR1681763

See PR Fix Info -> Root Cause for details: VxLAN Terminal End Point (VTEP Nodes) are expected to be reachable over the Data Path. If a route to the VTEP is resolved over the management interface of the switch/router, based on the current route tables, this is usually a configuration issue and may lead to further problems. PR1688296

For leaves of data type ieeefloat32, the value will be encoded in bytes while being streamed to collector. The value contained in such leaves may not be completely accurate. PR1690598

FIPS mode is not supported in this release for SRXSME devices. PR1697999

When subscribing to sensor paths "/junos/system/linecard/packet/usage/", "/junos/services/label-switched-path/usage/" or other line card (PFE) sensor paths in gNMI subscription mode, packet drops may be seen in the CLI command "show network-agent statistics gnmi detail" output. The collector output may also contain missing sequence numbers. For example, the sequence number output may be 0, 3, 6, 9, 12, etc. instead of 0, 1, 2, 3, 4, etc. PR1703418

In Chassisd, Jvision thread takes more time in streaming of jvision packets because of volume of data and number of sensors involved with this daemon. Jvision thread engaged for more time to process streaming events caused Chassisd master thread to lose receive/send keepalive messages to/from other RE, which eventually was causing automatic RE switchover in most of the cases. To avoid this, fix done for exporting small payload jvision packets (formation of which takes less time) and deferring jvision thread more in an interval, to allow chassisd master thread to process high-priority hello/keep-alive messages. This means now, more number of packets is sent in one reporting interval and with larger spread (earlier same amount of data was sent with 2 or 3 packets of higher payload size, and 100ms of deferring time for jvision thread. This behaviour is increasing KPI-2 but lowering KPI-1 (payload size). It is not possible to back out changes done to solve keep-alive message loss issue. Hence we will have to keep Chassisd as an exception, when we measure/report KPI-2 values. Jvision in Chassisd has to give more priority/time to process keep-alive messages than sending of jvision packets. Hence delay between jvision packets are more. PR1706300

Current stack and display is correctly set to 128 ports that is qualified on all MX10K8 line cards. PR1706376

On the MX104 platform, the Wrong threshold-temperature is displayed. PR1713788

fec-codeword-rate data with render type decimal64 is rendered as string in grpc python decoder.PR1717520

With no-reduced-srh configured, MX304 removes the last SID value from the SRH. Expectation is Last SID should be retained in SRH when "no-reduced-srh" is configured. There is no impact to the traffic. Traffic flow fine, since the "SEGMENT-LIST" and "LAST ENTRY" are encoded properly in the packet. PR1721404

In some srv6 scenarios, with no-reduced-srh configured, next header in SRH is not set and packets may be dropped as invalid hop option. PR1721429

On the Junos QFX5200 platform, sometimes upon restarting the device the 100G link will not come up and will remain down, impacting the traffic flowing through it. PR1725116

On Junos MX platforms, to enable Enhanced Subscriber Management feature without 'max-db-size' configuration on router >=32GB DRAM(Dynamic Random Access Memory), router needs to be rebooted only once instead of rebooting twice. PR1732216

There is no functional impact but the previously installed JSU will show up even though it is deleted during major upgrade. This PR will fix that issue. Workaround is to remove /packages/sets/active/junos-version file. PR1732878

Errors might be seen on ephemeral commit during ISSU.PR1679645

BGP LU statistics does not report correct statistics when sharding is enabled. This is not specific to BGP CT feature of this RLI. PR1684238

This issue is seen with only evo and not seen Junos. Its seen in a combination of Rsvp and ISIS. Stats is getting incremented. PR1700063

Show route advertising-protocol bgp reporting NextHop self rather than IP in the configured policy-statement for next-hop. Behavior change observed after JUNOS upgrade from 18.4 to 20.4. #set policy-options policy-statement set-NH-MX term to-PP-All then next-hop 20.20.20.1 show route advertising-protocol bgp 10.10.10.10 test.inet.0: 5 destinations, 5 routes (5 active, 0 holddown, 0 hidden) Prefix Nexthop MED Lclpref AS path * 10.0.0.0/31 Self 65000 I The CLI output for Nexthop reported Self rather than IP address 20.20.20.1. PR1712527

On all Junos and Junos Evolved platforms with TI-LFA (Topology-Independent Loop-Free Alternate) feature enabled, when IP address is removed from one interface and is assigned to another interface in the same commit, the rpd process crashes affecting routing control plane. PR1723172

Tunnel debugging configuration is not synchronized to the backup node. It needs to be configured again after RG0 failover. PR1450393

On all Junos and Junos Evolved platforms, when OSPF inter-area is configured with segmented provider-tunnel and master undergoes MBB(make-before-break), the multicast route entry on backup router will not have the tunnel name synced with master. PR1710323