| PR Number |
Synopsis |
Category: HW Board, FPGA, CPLD issues |
| 1491803 |
Failed to start MGD initialization of schema and database while upgrade from Junos OS Evolved 19.3R2.1 to 19.4-202002240046.0-EVO ]: mgd-init service fails due to clksyncd failure.
Product-Group=evo |
A QFX5220 may fail during an upgrade from 19.3R2.1 to 19.4-202002240046.0-EVO or later release due to "clksyncd" failure. |
| PR Number |
Synopsis |
Category: "agentd" software daemon |
| 1470872 |
xmlproxyd binding queue complete-deleted anomalies seen for the type net::juniper::config::services::agent.
Product-Group=evo |
This can happen in negative test cases. When customer installs a yang package then configures a sensor and then deletes the yang package before deleting the sensor. |
| PR Number |
Synopsis |
Category: PTX10003 Platform Resiliency |
| 1499291 |
CRC errors on Interface may result in Cmerror and ASIC errors in logs.
Product-Group=evo |
CRC errors on an interface might result in Cmerror and ASIC errors in logs. |
| PR Number |
Synopsis |
Category: software upgrade infra issues |
| 1495655 |
Block software upgrade when image that is not compatible with the platform is used.
Product-Group=evo |
Junos Evolved software upgrade process does not detect an Image's incompatibility for a platform. For example, we don't report an incompatibility when trying to use an image for PTX10003-EVO on a PTX10008-EVO and vice versa. The system ends up with an incompatible image and needs to boot from snapshot to recover. |
| PR Number |
Synopsis |
Category: EVO L2 Control Plane PRs |
| 1457825 |
The switch-options configuration not available on lean rpd images.
Product-Group=evo |
switch-options config not available on lean RPD images |
| PR Number |
Synopsis |
Category: EVO linux defects & enhancement requests |
| 1482363 |
EVO: Telnet login related issue with Template (TACACS & Radius)
Product-Group=evo |
When a TACACS user or a RADIUS user logs in through Telnet, the username displayed in login prompt, show cli authorization output, show system users output, and accounting logs is template username and not the actual username of the logged-in user. The issue here is a display issue where the username displayed will be the template username instead of the actual logged -in username, but there is no functional issue. The problem is specific to Telnet and when the user logs in through SSH, the actual logged-in username is displayed. |
| PR Number |
Synopsis |
Category: Configd, ffp issues |
| 1492765 |
Delay in deleting PrefixListConfig objects after switching between base config and scale config multiple times
Product-Group=evo |
When changing the configuration between scale configuration and configuration containing no filter or apply-path, sometimes there is a delay in deleting the apply-path addresses. |
| PR Number |
Synopsis |
Category: Express PFE FW Features |
| 1492708 |
[firewall] [filter_installation] EVO: Filter having apply-patha as match condition is not hit correctly after attaching to ae interface
Product-Group=evo |
When ipv6 filter is applied to ae interface after an empty config and the filter has prefix-list match-condition having apply-path, transit traffic does not pass through |
| 1495118 |
PTX10003: Offline/Online of FPCs is not fully supported
Product-Group=evo |
Offline/Online FPC feature is not fully supported on PTX10003 series. We can still do offline, but for bringing FPC or the associated ports back online requires a reboot of system. It is recommended we do not use FPC offline/online this feature is completely supported. It was also seen in some cases with scale configurations and heavy churn in system, off-lining an FPC were disabling packet forwarding on other FPCs as well. To recover, system reboot the system. |
| PR Number |
Synopsis |
Category: Resource Reservation Protocol |
| 1501860 |
When changing the config from WECMP<-->ECMP, traffic is not taking mpls route
Product-Group=evo |
on the fly change of wecmp to ecmp and vice versa will not take full effect for all routes using the LSPs until all those LSPs currently signaled with non zero bandwidth have gone through at least one of the MBB/clear/disable event, the traffic distribution will not be as expected until all those LSPs finished such transition. the reason is because without any of those triggers, the lsp will not automatically update the its nexthop based on wecmp configuration, nexthop content of each participation LSP will affect the traffic distribution among ECMP paths. the WECMP distribution ratio will not be as expected until the adjustment of all LSPs are finished. |
| PR Number |
Synopsis |
Category: Express PFE L2 fwding Features on ZX platforms |
| 1493132 |
Brackla L2: Burst of learning events causes linecard Major errors
Product-Group=evo |
Learnt MAC's on brackla are transported as Mac learning packets i.e. MLP's towards PFE CPU. When there is burst of learning events, we can land into scenarios where these packets use more bandwidth than pfe's lcpu interfaces supports. We debugged this with ASIC team and they have confirmed that these errors will appear when more traffic is enqueued towards LCPU than it can support. Once MAC's are learnt these errors stop. Also there is no functional impact as such of these errors because TOE ucode would retry with MAC learning packets, if these are dropped anywhere in packet path. |
| PR Number |
Synopsis |
Category: Express ZX PFE L3 Features |
| 1509694 |
Brackla load balancing - Sometimes Ipv6 src/dst address hash fields are not set by default
Product-Group=evo |
IPv4 src/dst and layer4 port(tcp/udp) is enabled by default(when boot up system with no related config knob configured). If user configured above know and delete the knob, the ipv6 hash-key fields may be disabled. If user want to make sure these fields are enabled, it need explicitly config the knob in cli configuration: config knob for explicitly enable ipv4 src/dst: set forwarding-options hash-key family inet layer-3 config knob for explicitly enable ipv4 layer4 port(tcp/udp): set forwarding-options hash-key family inet layer-4 |