Open Issues

In EVPN multihomed (ESI) deployments without core isolation, a temporary traffic duplication and brief traffic interruption could occur during a spine reboot. This behavior was caused by timing gaps between EVPN multihoming synchronization, DF election, and local-bias filter installation, which allowed traffic to be forwarded before multihoming state was fully converged. PR1879311

USB disks with Junos OS Evolved images from Junos OS Evolved 23.4R2 onwards might not be detectable by Windows. They still have valid images, and can be used for Junos OS Evolved installs. The only issue is that new images cannot be installed on these USB disks because Windows no longer recognizes these USB drives.PR1819846

Link with the DAC (SFP56-50G-DAC-3M ) comes up with 25G default speed configuration when switch is rebooted. Hence, When 50G speed config is applied , peer side sees the link flap. PR1836697

Insufficient headroom cells are allocated per Priority-Group(PG) when cable-length is configured <100mts(default length) under 'congestion-notification-profile' configuration and entire traffic frame size is <128B. Because of this one can observe the Ingress drops on the Port/Interface. PR1846357

Hitting with input-errors and framing-errors on channelizing QSFP56-DD-400GBASE-DR4. PR1848109

During USB boot with a fully populated system, error messages were observed in the boot logs indicating IRQ was not handled. This has no functional impact, as the driver is not used on the QFX5241-32/64 platforms. PR1868237

Graceful JOJI of FPC is the recommended way instead of ungraceful.PR1874712

With a back-to-back connection using 800G ZR* optics, if one end is configured with native speed (default) and other end is configured wit 8x100G mode, then the expectation is that link is down on both ends because of the configuration mismatch between the ends; but one of the two ends shows as up. The same behaviour is presented by 400G ZR* optics also. PR1884559

QFX5241-64OD:QFX5240-64OD: After configuring igmp snooping, group_count is not as expected since groups show in IRBs. PR1888392

PTX10008:Verification of Packet Forwarding Engine BFD sessions failed after switchover from primary Routing Engine to backup Routing Engine on the device, observing count does not match after giving request pfe execute command show frr-session-db 1 PR1889847

QFX5130 macsec: on macsec enabled interfaces, after evo-pfemand restart with CLI restart evo-pfemand, ping might not work sometimes.PR1891444

A little higher convergence (~1-2 seconds) seen when following conditions are met. 1. A MH leaf is connected to more than 1 spine. 2. In 2 spines scenario, initially one of uplinks is down on the leaf. 3. BFD packet must hash to this path. 4. Wait for (4-6 seconds) and then make second uplink down from another spine (now leaf is isolated). 5. Convergence of ~1-2 seconds is seen. Higher convergence is not seen when 1. Both spine uplinks are down at the same time. 2. when remote leaf is rebooted.PR1894826

Convergence of 5-13 seconds seen when uplinks from remote leafs to core are brought up. This is for type5 tunneled traffic only.PR1895977

NETCONF server reply to get operation pointed to /components/component[name=""]/state/will not include any OPTICAL_CHANNEL type component data. PR1897729

QFX5700: 16x100G PIC: Commit check added when 40G and 100G xcvrs are put to share a Broadcom serdes core. PR1900911

Whenever picd process restarts, the interface et-0/0/1 might go down and will not come up even with interface disable-enable. PR1898893

On QFX5130 platforms, in VXLAN topologies, Symmetric ECMP hashing might not be enforced when resilient hashing is concurrently enabled. PR1907397

This issue is seen intermittently on few devices. Probably due to some other kernel modules/processes using the i2c lines for these SFP voltage controllers. Impact: During firmware upgrade of sfpg1/g2/g3. once in a few iterations there will be write failures, which is expected in situations where another driver is interacting with device on same i2c line. Retry will make it work. Recovery: There is no other impact on the device except upgrade failure, which is rare if customer wants to upgrade once. in case encountered, retry will solve the issue.PR1908672

On a QFX5241/QFX5240 platform, changing the MTU value of an interface does not take effect when the UFT profile is also being changed in the same commit. The MTU configuration should be applied in a separate commit.PR1914797

On Junos OS Evolved QFX5000 platforms classifier and rewrite bind supported either under unit 0 or unit * , but not together. Commit check enforced to reject such configuration. PR1920633

On QFX5230-64CD platform, 400G DAC cable of 2.5m length and 4X100G DAC BO may not link up with some peer devices. This issue is not seen with all peer devices. If this happens, please replace the 2.5m cable with a 1m DAC cable or use supported 400G Optics instead. PR1747315