Search

Print this pagePrint this page

Email this topicEmail this topic

View PDFView PDF

Download EPUBDownload EPUB

Open Issues

The following are new open issues for supported features found in ExtremeXOS 21.1.

expand icon

Open Issues, Platform-Specific, and Feature Change Requests (CRs)

CR Number Description
General
xos0062584 The command configure web http access-profile is missing the option access-profile.
xos0062966 When rendezvous point receives (*, G) join, and it has (S, G, RPT) entry, the entry should be converted to (S, G), and (S, G) join should be sent upstream. However, sometimes wrong assert is triggered and this new entry is dropped after a minute, resulting in complete traffic loss.
Note: Disable, and then re-enable PIM.
xos0063346 With multiple (greater than two) VRRP instances and host-mobility enabled, FDB flush sent during topology change from other L2 protocols does not occur.
xos0063247 Duplicate packets occur after MLAG recovery.
xos0063396 While enabling 1G links, unexpected link-down PDU is received in EAPS master.
xos0048715 IPv6 ECMP works for hardware-forwarded traffic, but does not work for slow-path traffic.
Note: Either use BFD to keep all router neighbors alive, or configure static neighbors and static FDB entries for all router neighbors. BFD is the preferred method.
xos0062399 IPv6 BFD session for OSPFv3 flaps after disabling, and then enabling VLAN port.
SummitStack
xos0062386 With BGPv6, after port flap or failovers, some peers go into idle state.
xos0061909 Creating an IPFIX mirroring instance to a monitor port, deleting the mirroring instance, and then recreating it again with to a different monitor port, causes the following error message (similar to the one below) to appear, and IPFIX mirroring does not work:
<Erro:HAL.Mirror.Error> Slot-1: Failed to create mirroring destination for slot 2, unit 9 Entry exists
Note: If the error appears in the log, disable and delete the mirror instance, and then add it back again. 
ExtremeSwitching X620 Series Switches
xos0062729 On ExtremeSwitching X620 series switches, for ports with Base-T SFP optics and explicitly configured at 1,000 speed, link comes up at peer end, but link stays down at local end after either rebooting, or removing, and then re-inserting optics.
Note: Manually set autoneg to "off" on the host side, which allows the link to be established.
xos0062636
Unexpected link switchover behavior occurs when exchanging copper and fiber cables on ExtremeSwitching X620 combo ports.
Note:
  • When 10G combo ports are used at 1G for redundancy between fiber and copper, then set the preferred medium to copper (configure ports port_list preferred-medium copper), otherwise sometimes the copper link might not come up.
  • When 10G combo ports are used at 10G for redundancy between fiber and copper, then set the preferred medium to fiber (configure ports port_list preferred-medium fiber), otherwise sometimes the copper link might not come up.
xos0062620 For ExtremeXOS 21.1, do not use copper DAC cables for stacking on ExtremeSwitching X620-16T switches.
Summit X670-G2 Series Switches
xos0063170 On Summit X670-G2 series switches, greater EAPS convergence time occurs with multiple VLANs (1,000 protected VLANs).
Summit X460-G2 Series Switches
xos0063811 Summit X460-G2 series switches with ExtremeXOS 15.6 through 21.1, have the following limitations for SyncE input reference frequency:
  • Network clock does not lock with input SyncE source port 52 (both at 10G and 1G speed) on all 48-port models (X460-G2-48t, 48x and 48p).
    Note: For SyncE input at 10G, avoid port 52.
  • When the 10G ports operate at 1G speed, the network clock does not lock. Models with Ethernet BASE-T or Power over Ethernet (PoE) ports may lock on initial configuration, but do not lock after a save and reboot.
    Note: For SyncE input at 1G, use a 1G port, not a 10G port.
BGP
xos0060641 When BGP is administratively shut down, it does not send notifications to peers.
xos0063778 If an applied BGP import policy is edited such that previously permitted routes are now denied, the BGP RIB (show bgp routes command) still shows the newly denied route(s) as active. The routing table is, however, updated correctly to reflect the new policy.
xos0063698 A BGP route is not replaced in the routing table by a new instance of the same prefix and length containing a different metric value. This condition can occur if an applied BGP import-policy file is edited to modify the route metric.
Published March 2016prev | next

Email this topicEmail this topic

Print this pagePrint this page