Open Issues

End-hosts might not communicate via Ethernet VPN with Virtual Extensible LAN encapsulation (EVPN-VxLAN) domain after Ethernet Segment Identifier (ESI) failover. This issue affects QFX5000 platforms only. Please refer to restoration steps when this issue is encountered. PR1584595

Modifying the I-ESI value is a traffic effecting event. If this must be done, then use the following steps to avoid this issue.

1. Deactivate interconnect stanza for the routing-instance in question.
2. Modify the I-ESI value.
3. Activate the interconnect stanza.

PR1600600

This problem happens only with translation VNI when mac is moved one from DC1 to DC2. VM move across DC where there is no translate VNI configuration in the interconnect works as designed. PR1610432

EVPN Local ESI Mac limit configuration might not become effective immediately when it has already learned remote MH Macs. Clear the Mac table from all MH PEs and configure the Mac limit over local ESI interfaces. PR1619299

User cannot change I-ESI on multi-homing GW nodes at the same time. There will be some gap in commit in two different GW nodes. During the gap, there will be different I-ESI on MH GW nodes, which will cause the loop for a short duration. Due to the loop, mac goes to duplicate mac detection state. To clear this duplicate macs, user has to issue clear duplicate mac detection. As a workaround, deactivate interconnect on multi-homing GW nodes before changing the I-ESI, commit, and then change I-ESI and activate interconnect. PR1624981

The following messages are seen during FTP:ftpd[14105]: bl_init: connect failed for `/var/run/blacklistd.sock' (No such file or directory). PR1315605

In QFX, when BUM traffic is sent on MCLAG, Mac entries are learnt on ICL interface as DLR when ICL is flapped as mac learning. This may cause the traffic loss with certain traffic flow. PR1639713

If request system zeroize does not trigger zero-touch provisioning, re-initiate the ZTP as a workaround. PR1529246

The DHCP client configuration is coming from two places, that is, AIU script and vsdk sandbox. The DHCP client configuration coming from AIU script has the serial Id in vendor id where as the default configuration from sandbox doesn't have it. There is no impact on functionality or service. PR1601504

The rsvp interface update threshold configuration syntax has changed between Junos OS Release 18.2X75-D435 and Junos OS Release 20.3X75-D10 to include curly braces around the threshold value. Upgrading and downgrading between these releases is not entirely automatic. The user must delete this stanza if configured before the downgrade and then manually reconfigure. PR1554744

The commit synchronize command fails because the kernel socket gets stuck. PR1027898

Arrival rates are not seen at system level when global-disable fpc is configured on QFX. PR1438367

When the DHCP relay mode is configured as no-snoop, we are observing the offer gets dropped due to incorrect asic programming. This issue only affects while running DHCP relay on EVPN/VXLAN environment. PR1530160

On a PTX Series router or QFX10000 switch with a third-generation FPC, an error message is displayed when the FPC goes online or offline. PR1322491

QFX10000:Source MAC and TTL values are not updated for routed multicast packets in EVPN-VXLAN. PR1346894

Backup Routing Engine might crash after GRES occurs continuously for more than 10 times. PR1348806

When VLAN is added as an action for changing the VLAN in both ingress and egress filters, the filter is not installed. PR1362609

Due to transient hardware condition, single-bit error (SBE) events are corrected and have no operational impact. Reporting of those events had been disabled to prevent alarms and possibly unnecessary hardware replacements. This change applies to all platforms using Hybrid Memory Controller (HMC). PR1384435

The unified ISSU is not supported on QFX5200 switches and fails from Junos OS Release 17.2X75-D43.2 to some target versions. Also, dcpfe crash might be seen. PR1438690

There is a timing issue during the sxe interface bring up (w.r.t i40e driver). This can be recovered by rebooting the complete board. PR1442249

In race condition, if a BGP route is resolved over the same prefix protocol next hop in a routing table that has routes of the prefix from different routing protocols, when the routes are flapping (firstly these routes are down and then up), the BGP route will be re-resolved, and then the rpd crash might be seen. PR1458595

The storm-control does not rate-limit ARP packets on QFX10000 although shutdown action works. PR1461958

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

On QFX5000/EX4600 Series platforms with "instance-import", deleting route which has "next-table" used might result in unexpected route next-hop. PR1477603

When running the show pfe filter hw filter-name filter name command, the command fails to retrieve the Packet Forwarding Engine programming details of the filter. PR1495712

On Junos PTX and QFX platforms, the Jflow service might not report the accurate throughput rate. This issue is seen when there is high sampled traffic rate with low flow cache hit ratio. PR1502645

On a fully scaled system where all the slices are utilized by different families of CLI filters, if you try to call delete for one family and add/change for another family with a higher number of filter terms which requires either expansion of the filter or creation of a new filter, the Packet Forwarding Engine fails to add the new filter as we are getting messages out of sequence that is, the add/change of filter is called earlier than the delete of another filter that will free up the slices. PR1512242

MSDP sessions might reset after a GRES reset even when nonstop routing state is synchronized and ready for switchover. PR1526679

PR 1530951: FIPS mode is not supported on 20.4R1-EVO. PR1530951

On QFX5100 devices not running the qfx-5e codes (non-TVP architecture), when an image with the Broadcom SDK upgrade (6.5.x) is installed, the CPU utilization might go up by around 5%. PR1534234

On QFX5000 serial platform, under Ethernet VPN and Virtual Extensible LAN (EVPN-VXLAN) spine-and-leaf environment, packets sent via VXLAN tunnel might lose Virtual Tunnel End Point (VTEP) IP address (source IP address (SIP)). Packets without SIP could cause memory leak. This issue rarely happens because it is difficult to be reproduced. PR1536895

"Socket to sflowd closed" error comes up when the ukern socket to sflowd daemon (server) is closed. The error is rectified by itself as the client successfully reestablishes the connection in the subsequent attempts. When these errors are consistent, it indicates a communication issue between sflowd and the sFlow running on the FPC. PR1538863

EVPN-VxLAN: vmcore files are seen on Master and backup Routing Engine of QFX10k8 with L2/L3 Multicast configuration. PR1539259

Broadcom has updated that BCMX calls are deprecated and needs to be replaced with BCM calls. PR1541159

100G AOC from Innolight does not comes up after multiple reboots.It recovers after interface enable/disable. PR1548525

Uusers cannot subscribe to any path that ends with "key". PR1553534

ISSU is not supported from releases below 20.4 to releases 20.4 and above. There is a major SDK upgrade from 6.3.2 to 6.5.16, due to which the Warm boot feature needed for ISSU is not supported by our vendor. PR1554915

5M DAC connected between QFX10002-60C and MX2010 doesn't link up. But with 1M and 3M DAC this interop works as expected. Also it is to be noted QFX10002-60C and ACX or Traffic generator the same 5M DAC works seamlessly. There seems to be certain SI or link level configuration on both QFX10002-60C and MX2010 which needs to be debugged with the help from HW and SI teams and resolved. PR1555955

In Release 20.2, some features will show up as a licensed feature. Customer would see alarms, commit warnings and show system license output as below. However, there would be no functional impact. user@QFX5120> show system license License usage:Licenses Licenses Licenses ExpiryFeature name used installed neededesi-lag 1 0 1 invalid. PR1558017

On QFX5120-48Y switches, amber LEDs are displayed for the fan modules after upgrading to Junos OS Release 20.2R1 and later. The LEDs are on continuously even if there is no fault in the fans. PR1558407

To avoid the additional interface flap , interface hold time needs to be configured . PR1562857

Starting in Junos OS Release 21.1R1, Junos will be shipping with python3 (python2 is no longer supported). In ZTP process, if a python script is being downloaded, please ensure the python script follows python3 syntax (there are certain changes between python2 and python3 syntax). Also, so far (that is, until 20.4R1), the python script had #!/usr/bin/python as the first line (ie the path of the python interpreter). The same needs to be changed to #!/usr/bin/python3 from Junos OS Release 21.1R1. PR1565069

The chassisd logs are flooded with "pic_create_ifname: 0/0/0 pic type F050 not supported" messages for every port that is connected. This will happen every few seconds. PR1566440

In mixed QFX5100, EX4300 VCF setup, duplicate traffic might be observed for some Layer 3 multicast traffic streams . PR1568152

On QFX5000, in EVPN_VXLAN deployment, BUM traffic replication over VTEP might send out more packets than expected. PR1570689

In EVPN/VXLAN scenario with OSPF configured over the IRB, OSPF sessions might not get established due to connectivity issues. PR1577183

Observed the partial traffic loss after disabling the protected link on R2 hence seeing the convergence delay for link-protection for PE1_P link. PR1579931

On QFX5100, while checking DHCP Smart Relay over IRB interfaces, the renew-ack's may not be seen in the dhcp client. PR1581025

In a fully loaded devices, at times, firewall programming was failing due to scaled prefix configuration with more than 64800 entries. However, this issue is not observed in development setup. PR1581767

When physical loopback is used and both the ports are with EP style in the same RSPAN VLAN it can lead to flooding. PR1581876

RPD core file is seen @rt_iflnh_set_nhid . Core file is due to assertion caused by failure of hbt_insert for nhid belonging to an ifl. It is seen that there is a duplicate entry present which causes the hbt_insert failure. PR1588128

The 'input ingress' and 'input egress' together for the same port will not work in mirroring with VXLAN encapsulation. PR1589854

On QFX/EX series switches with Broadcom chip as Packet Forwarding Engine (PFE), if IS-IS is enabled on an integrated routing and bridging (IRB) interface and the maximum transmission unit (MTU) size of the IRB interface is configured with a value great than 1496 bytes, the IS-IS hello (IIH) PDUs with jumbo frame size (that is, greater than 1496 bytes) might be dropped and not sent to the IS-IS neighbors. The following is the product list of QFX/EX series switches with Broadcom chip as Packet Forwarding Engine. QFX5100/QFX5110/QFX5120/QFX5130/QFX5200/QFX5210/QFX5220 EX2300/EX3400/EX4300/EX4600/EX4650 PR1595823

On the QFX10000 series platforms, the process dcpfe/FPC crash might be observed in boot time if rebooting the devices. PR1597479

There is a remote possibility that during many reboots, the Junos VM goes into a state where NMI is needed to continue the reboot. There is no workaround for this and a subsequent reboot does not seem to hit this issue. PR1601867

During reboot in certain instances device gets into a state where Junos virtual machine hangs until the NMI is triggered and reboots fully. PR1602360

Chassisd generates "Cannot read hw.chassis.startup_time value: m" every 5 seconds on qfx10008. PR1603588

On QFX5100, optical power is seen after detached and attached QSFP on disable interface. PR1606003

In QFX10002-60C under mac statistics "output-mac-control-frames" and "output-mac-pause-frames" does not increment. PR1610745

QFX10002/QFX10008/QFX10016: As per design change, devices have max token allocation limit of 104k combined for l2/l3/BUM routes. These tokens are used as egress nexthop IDs for route lookup in HW. Out of 104k tokens, 8k tokens are reserved for maximum VXLAN tunnels and 16k tokens are reserved for IRB interfaces, and locally learned ARP/NDP Nexthops also require tokens to install in HW which will be allocated from free pool of tokens left after the reservations. So, total tokens available tokens in free pool after reserving [8k(vxlan tunnels) + 16k(IRBs)] is 80k tokens. Thus, maximum of 80k local ARP entries can be supported on QFX10k (Elit/Ultimate) devices. QFX10002-60c: As per current design, device have max token allocation limit of 96k combined for l2/l3/BUM routes. These tokens are used as egress nexthop IDs for route lookup in HW. Out of 96k tokens, 8k tokens are reserved for maximum VXLAN tunnels and 10k tokens are reserved for IRB interfaces, and locally learned ARP/NDP Nexthops also require tokens to install in HW which will be allocated from free pool of tokens left after the reservations. So, total tokens available tokens in free pool after reserving [8k(vxlan tunnels) + 10k(IRBs)] is 78k tokens. Thus, maximum of 80k local ARP entries can be supported on QFX10k (Elit/Ultimate) devices. PR1616224

In "Remote Mirroring with VXLAN Encapsulation" Physical/Softloopback ports are used to achieve the VXLAN encapsulation of the mirrored traffic. When mirrored traffic packets are of NS/ARP these packets are sent to the CPU and are re-injected back to RSPAN VLAN, which will cause the LOOP between the loop-backed ports and in turn sent to the remote collector. PR1631618

On QFX platforms, in Q-in-Q scenario when 'input-native-vlan-push' statement is disabled inner Virtual LAN (VLAN) might be stripped off if it matches the outer VLAN. Traffic drop might be observed in this scenario. PR1631771

chassisd crashed after configuring chassis disk-partition. The cli command is not supported by QFX10008 and QFX10016 platforms. Hence disabling the same. PR1635812

Child link is not coming up after loaded the lag configuration. PR1646495

Multicast traffic is overusing the switch core when igmp-snooping is removed. The MCSNOOPD will generate core files due to the changes in mrouter interfaces and routes. PR1569436

On QFX10002 platforms, the multi-hop BFD session might flap if collecting RSI or some other outputs (such as show interface or configuration). It is caused by the missing BFD packets because the PPMAN thread is not scheduled within the BFD timers which are 300 milliseconds with a multiplier of 3. PR1589765

When the statement accept-remote-source under PIM is removed, the PIM SG entries might not be updated with the correct RPF. Clearing of the states would take care of the issue. This is day-1 behavior. PR1593283

On certain conditions where underlay session flapped due to interface down event, it may possible to see the flap in overlay bfd session due to delay in route install to reach the packet to the neighbor. PR1618118