Skylight sensor control 20.11.1 Release Notes

These release notes cover the requirements, new features, and changes, for this release of the SkylightTM sensor: control firmware version 20.11.1. Please read all the notes prior to installing this firmware version.

Organization

This manual includes the following sections:
Requirements
This section provides requirements for Firmware Version 20.11.1
Skylight sensor: control 20.11.1 Lifecycle
This section provides the release lifecycles for Skylight sensor: control 20.11.1

Requirements

This firmware version applies to the following products from Accedian:

Skylight sensor: control was tested using the Skylight Appliance version 1.4.

Skylight sensor: control 20.11.1 requires Skylight orchestrator 20.05 as its minimum. Customers using SD Vision must first upgrade to Skylight orchestrator 20.05 before upgrading to Skylight sensor: control 20.05.

Important note

In Skylight sensor: control 20.11.1, any module upgrades from firmware versions previous to VCX 2.2 FWSuite (FWSUITE_VCX_2.2_10190) have been blocked, in order to protect against a potential complete loss of connectivity/functionality that cannot be recovered. In order to upgrade a module from an older FWSuite version, an older version of Skylight sensor: control must be used to first upgrade the module to FWSUITE_VCX_2.2_10190:

Skylight sensor: control 20.11.1

• VCX_20.11.1_23836
• FWSUITE_VCX_20.11.1_15890

Skylight sensor: control upgrade consideration

When upgrading the Skylight sensor: control software, the module's firmware is not upgraded automatically. It is not recommended to operate Skylight sensor: control with a different module firmware version since the functionality is limited to management. The management includes:

• Module discovery
• Connect to module and have a management Link state
• Upgrade module firmware version using feature suite
• Update module interface (IP address, VLAN, and DHCP)

It is not recommended to change any other module configuration when operating with a different firmware version. Changing the configuration could result in unknown behavior. A factory reset using AMD could be required in some cases. In a future release, the Skylight sensor: control software will prevent changing configuration for modules running a different firmware version.
For cases where all modules cannot be upgraded at the same time, it is recommended to run different Skylight sensor: control instances with different software versions. Modules to be upgraded should be moved between Skylight sensor: controls.
When downgrading Skylight sensor: control software, the modules firmware version shall also be downgraded. The downgrade process shall follow these steps in order to successfully downgrade Skylight sensor: control software and modules firmware. Note that downgrades are not recommended (see CAUTION above).

Scenario 1: downgrade using Skylight sensor: control rollback

• The modules firmware must be downgraded first using the newest Skylight sensor: control software.
• Perform Skylight sensor: control rollback.

Scenario 2: downgrade using a different Skylight sensor: control instance

• The modules firmware must be downgraded first using the newest Skylight sensor: control software.
• Delete modules from the newest Skylight sensor: control instance.
• Discover the modules using the older Skylight sensor: control instances.

Firmware Version 20.11.1_23836 (2021-02-05)

System capabilities

The Skylight sensor: control offers the following system capabilities:

Operational Considerations

Important Notes

• When using hypervisor functions that involve dynamic load balancing of guest virtual appliances in a hypervisor cluster (.i.e. VMWare DRS), the balancing policy should limit rebalancing operations to be done on virtual appliance startup and/or hypervisor host failures. The dynamic load balancing must follow the engineering guidelines for dedicated resources allocation. The user should also set MAC preservation. This is mandatory when using local licenses as the license is coupled with the MAC.
• If multiple discovery methods are being used, it is possibly that they will discover the same remote device. This can cause some confusion regarding the IP address being used to manage the unit, especially if a specific IP address is expected to be used for a specific remote device. It is important to note that the first discovery method used to discover the remote device is the one that will be used for management. For example, if DHCP advertisement is used at the time of an IP Agnostic discovery, then the IP address of that first method discovering the device is the one that will be used. It is good practice to use only one discovery method for a specific remote device.

• When user completes an upgrade to the latest build without configuration reset, they will get an error "CSRF Violation" when uploading the import files for the first time.
• Press Ctrl + F5 to refresh the page.

• The 10G SFP compute link can take 30 seconds to come up.
• The 10G SFP compute reboot time can take up to 60 seconds.

• SyncE:
• Shall be use with no force link up disable and ESMC enabled.
• Long term holdover not supported.

• Internet Explorer 11 is no longer supported. The browser does not support newer technology and does not always work properly.
• Loopback usage in the second-generation modules is limited to one loopback on both ports or two loopbacks on a unique port. Using two loopbacks on both ports at the same time will be removed in a future release.
• The virtual machine MTU must be configured to a value greater than 1526 bytes in order to generate 1500-byte NFV probes over a Q-in-Q interface.
• The virtual machine disk must be deployed using thin mode with ESXi in order to limit the storage to the configured size. Otherwise, the maximum configuration size will be reserved on the host.

• The following table provides the traffic downtime associated with the upgrade of a Skylight sensor: control 20.11 release firmware suite. All values are expressed in seconds.

  ---

  Note: The switching time (i.e., jump) between the PMON and TGEN firmware loads is equivalent to the FPGA downtime (third row below).

  ---

  Down Time	Skylight sensor: SFP compute 1G Copper	Skylight sensor: SFP compute 1G Optical	Skylight sensor: module 1G Copper	Skylight sensor: module 1G Combo	Skylight sensor: module 10G	Skylight sensor: module 1G	Nano 10G
  MCU	4.30	3.92	5.08	2.14	3.97	4.28	2.27
  Baseload	N/A	N/A	N/A	N/A	N/A	N/A	N/A
  FPGA	3.48	2.30	3.60	1.05	0.96	1.24	17.88
  Total	7.78	6.22	8.68	3.19	4.93	5.52	20.15

  The traffic downtime values shown above were calculated following firmware upgrade tests performed with Accedian Skylight performance elements acting as the host devices. Traffic downtime can vary from one host device type/model to another. For example, downtime measurements using a Cisco 901 as the host device gave the following results:

  • Skylight sensor: SFP compute 1G copper Skylight sensor: module downtime: 13.0
  • Skylight sensor: SFP compute 1G optical Skylight sensor: module downtime: 7.4

• SyncE clock transparency on Skylight sensor: SFP compute 1G copper Performance Modules may not work if the mastership is misconfigured.
• An XML file with the required Skylight sensor: control virtual machine hardware information has been provided with this release to ease deployment of the Skylight sensor: control on a KVM Hypervisor. Offered in libvirt-compatible format, this file can be used with any third-party tool that supports this format such as the virsh command line utility. For additional information, refer to libvirt.org.
• Prior to deploying the KVM image, you must configure your host networking settings to map to the Skylight sensor: control network interfaces.

• Flowbroker file transfer using FTPS:
• Ensure the FTPS server allows session re-creation. Otherwise, the file transfer aborts and the file is empty.

• SAT RFC-2544:
• Due to the number of traffic filters available per remote device, SAT RFC-2544 Layer-3 (IP) tests using multiple (i.e., up to four) flows must use the same UDP source and destination ports on all flows, otherwise one of the flows will not function properly.
• Attempts by the SAT RFC-2544 traffic generator to send two flows (Layer-3 packets with two distinct IP address targets) may occasionally fail. If this happens, simply restart the test.

• System Alarms:
• The threshold period cannot be defined for raised or cleared alarms.
• As the system reboots, some Loss of Connectivity alarms may be raised for remote devices that are configured in the system, but not yet linked to the Skylight sensor: control instance. These alarms are cleared when the remote devices are linked again.
• No alarm hierarchy mechanism has been implemented in the Skylight sensor: control. As such, no alarms are filtered if a higher-priority alarm is raised.
• Remote Devices:
• The remote device will be deleted if you change the port used for managing the remote device.
• Due to the ageing mechanism used by the remote device inventory, a remote device may still appear in the inventory once it has been removed, depending on the discovery period.
• A Domain ID cannot be specified when creating a remote device discovery instance using the ACP-Layer2 method. The Domain ID is automatically set to Default Domain.
• When deleting a remote device, the link between the Skylight sensor: control and the remote device must be properly closed before the same device can be added again. Remote devices added before the closure process has completed are not recognized by the Skylight sensor: control. In such cases, simply allow a few seconds for the closure process to complete before trying again.
• Since the Accedian ACP Layer-2 protocol is used to discover remote devices, its discovery messages may reconfigure the Auto interface of Accedian units (like Skylight element: TE, as well as NE and CE Skylight performance elements) running legacy firmware such as v4.9.x or older.
• When discovering more than 500 remote devices, it is strongly recommended to perform the discovery process, at most, once every 60 seconds. The three-second discovery feature is CPU intensive.
• The maximum permitted number of daisy-chained remote devices is 255.

• Time Synchronization:
• Only NTP client instances are supported by this product: NTP server instances are not supported. The NTP client presents certain limitations compared to other Accedian products.
• he date CLI command does not reject invalid dates.

• CFM second-generation Skylight sensor: module:
• The Skylight sensor: control manages CFM in either point-to-point or E-LAN topologies. These CFM messages are handled through an NFV tunnel established between a Skylight sensor: control.

• The DLA feature may require up to five minutes per remote device to update the loads contained on the remote devices that are linked to an instance of the Skylight sensor: control.
• If a Skylight sensor: module 1G has a combo port with one active port and one inactive port, both links will be shown as “up” because a remote partner is linked to each of them. The inactive port maintains an “up” link status as a way to achieve faster media selection.
• Skylight sensor: SFP compute 1G and Skylight sensor: module 1G cannot loop back any TCP frames addressed to them (device primary IP).
• A Skylight sensor: control’s Node ID can be edited after the Probe agent has been disabled. For this reason, a nodeid edit command must be preceded by the agent-server disable command, and followed by the agent-server enable command.

• Each instance of the Skylight sensor: control will extract the value of the following dynamic settings stored on the Skylight sensor: module:
• TWAMP stateless reflection state (enable/disable) and UDP port
• Default TWAMP stateful reflection state (enable/disable) and UDP port
• ETH-DMM reflection state (enable/disable)
• SyncE state (enable/disable) and clock source selection and QL state (enable/disable)
• LLDP enable state (enable/disable) and rate
• Any port PHY related settings
• Any port SFP related settings, such as:
• Laser state (enable/disable)
• Force Link Up (enable/disable) with timeout period

Skylight sensor: control 20.11.1 Lifecycle

This section lists the planned lifecycle dates of this software release. See the table below outlining the following milestones:

Import status is cleared automatically on destination Skylight sensor: control during rehoming of over 50 devices
The import status is cleared automatically after rehoming approximately 60 devices.
Workaround: Rehoming function should be done in orchestrator using batches of 50 devices at a time.

Support for regulator PL, DMM and SLM with pmcollect
Support added for Regulator PL, DMM and SLM with pmcollect.

**SFP compute 10G SyncE partially compliant **

• The SFP compute 10G, partially compliant to SyncE G.8262.1. clause 11.1, sometime fails; the phase jump is exceeding the maximum of 10ns specified in the clause. The phase jump measured is around 15 ns.
• Not compliant to G.8262.1 clause 10. The noise transfer should cut at 3 Hz and the actual noise transfer cut is at a frequency higher than 100 Hz.

PMCollect: CSV generation time is not correct after collection period change
After collection period is changed, the next file is still generated with the previous period. However records in files are correct. This behavior only happens for the first interval/period, and after this, the file generation time is correct.
Workaround: None.

“Time-of-day synchronization” alarm
When using sensor: control for PM sessions, and ARTS is not disabled, there is an alarm related with synchronization. This alarm should not be enabled if ARTS is disabled, but be raised only when ARTS is enabled and the ARTS is not synchronized.
Workaround: Disable the alarm when ARTS is disabled.

The import status is cleared while the import process is still happening on many devices
When importing more than 250 remote devices, the import status is cleared while the process is still occurring on many devices.
Workaround: Disable the auto update on Skylight sensor: control when doing the rehoming. Then the user will have to upgrade the device.

Duplicated data when streaming Flowmeter in 5 seconds interval
Flowmeter CSV file with period 5s is being duplicated 5 times.
Workaround: None.

RFC2544 TestSuite Throughput test Layer-3 fails and the packets reported have CRC errors on 10G remote devices
Sending packets in Layer-3 with dual VLAN causes to the RFC2544 testsuite test to fail and CRC errors are observed approximately as the same amount of the received packets (on NID).
Workaround: None

**ERROR: Bad Hostname, when using “” underscore **
If changing the hostname and using an underscore, in version 20.11 of Skylight sensor: control, a “bad hostname” error is received.
Workaround: Do not use an underscore (“”).

NFV PM using a 10G NIC card on sensor control can create packet loss
When using a 10G NIC card on sensor control for NFV PM, it is possible to have packet loss. The packet is caused by TCP burst faster than 1.25 Gbps. The TCP connection in the FPGA cannot go faster than 1.25 Gbps on 10G SFP compute and Ant 10G.
Workaround: Use a 1G NIC card or configure the NIC card to limit the TCP burst below 1.25 Gbps.

Flowmeter in wlic after VCX name change
Flowmeter sessions went into WLIC status and no CSV files were sent to Orchestrator starting at the time the time the name was changed. After 12 hours the sessions became active with no intervention.
Workaround: None. Wait and WLIC will chage to ACT, no intervention needed.

IPv6 address-add on existing IPv4 interface hangs the sensor: control
When trying to add IPv6 address to the existing IPv4 VLAN interface, it was noted that the Skylight sensor: control would hang, resulting in no response over WEB GUI, SSH or virsh console. The only way to recover this situation was to shut the system off from the hypervisor and start it again. Then it was noted that the IPv6 address was no longer there after the reboot.
Workaround:

• stop all PM sessions from the interface that is subject to IPv6 address add
• add IPv6 address
• start all PM sessions
• easiest way is to stop agent server, add IPv6 address, start agent-server

Web help missing parameter information for Exclude VLAN Size
Parameter information for the exclude VLAN size check box is missing in the following two locations:
Sat>RFC2544>Generator>Configuration, and
Sat>RFC2544>Testsuite>Configuration
Workaround: This issue will be addressed in the next release.

VCA-VLAN service VLAN must have same TPID when containing a tunnel
When the VLAN Stack Size >= 2 and Use-Tunnel is true, the VLANID assigned to the tunnel may be used by different service VLANs. When this happens, the TPID assigned to the individual service VLANs must be the same.

ARTS feature stays enable even after set to disabled
If ARTS enabled in concurrency with any VNID actuation protocol, the measurement will be invalid. ARTS needs to be deactivated to run VNID PM, if ARTS was prior activated, we need to deactivate ARTS and reboot the device, to reset the free running clock.

CPU utilization differs between Web UI and debug shell top
The CPU usage does not calculate the load average the same way that top is calculating it. The CPU usage should be used only for a CPU usage indication and does not reflect if the system is overloaded or not.

VCX unitid can change on new instance
The nodeid is built from eth1, so if only one port in VCX is spawned, the nodeid is build using dummy MAC address. If a second interface is added afterward, the nodeid will change.
Workaround: Create the two ports in advance in order to keep the unitid unchanged.

NFV tunnel does not connect when changing primary IP address of a device with static entry
When changing the primary device IP address (and the remote device is managed using static entry), the static entry destination address does not get updated automatically. The remote device management is moved automatically to the new primary IP address. The user must change the remote device static entry IP address with the new primary IP address. The NFV tunnel will not be connected until the static entry is updated.

SDLS does not release Probe Session license of Packet Loss after upgrade
In VCX 2.6 the Packet Loss and SLM are free of license. When upgrading from 2.5 to 2.6 the PM session licenses are not freed automatically.
Workaround: The licenses are re-synchronized every 24 hours. Therefore, after 24 hours the license will be freed.

VCX unresponsive when generating PM session using SDM on VCX local port
The VCX becomes unresponsive when generating PM session using SkyLIGHT Director Manager on VCX local port and then deleting the local port.
Workaround: The VCX must be rebooted and the SkyLIGHT Director Manager session on local port must be deleted and recreated.

History CSV files missing time stamps when scheduling period is 60 min and
Collection period is 1 min
When filling is not enabled, the history CSV file can contain a maximum of 32 reporting periods. Configuring a scheduling period of 60 minutes with a reporting period of one minute is an invalid configuration. The CSV file will contain only 32 entries.
Workaround: Configure the scheduling and reporting period to a limit 30 entries in the CSV file.

No sanity checking for filter for TWAMP stateful reflection
The filter for TWAMP reflection should contain a minimum of one UDP port in order to perform the reflection on TWAMP traffic only. Using a filter like catchall is equivalent to performing a full loopback on traffic and can cause network issues.

Bandwidth Regulator configuration disappears from WEB UI
Upon creation of 16 Bandwidth Regulators on a single device, the Bandwidth Regulator configuration information disappears from the WEB UI. To prevent this issue, the BW Regulator Name should be limited to seven characters.

SOAM libray V1 keeps sending CCM even if the NFV tunnel is down
The SOAM library V1 will keep sending CCM even if the NFV tunnel is down.

Not all created ETH-DM sessions statistics are displayed in VCX
Not all “creates ETH-DM sessions statistics” are displayed in VCX. Only the first 850 are displayed.

SDM cannot resolve VCX when deleting ptadmin user and recreating it
SDM cannot resolve VCX after deleting ptadmin and recreating it. The new ptadmin account does not have the same home path.
Workaround: Need to export the configuration, add the original ptadmin and import the new configuration.

Cannot create route with different netmask
Cannot create route with different netmask using the same IPv6 destination address.

MEP alarm and MEP status show different results after device reboot
Sometimes the MEP alarm is raised but the MEP status shows no alarms after a device reboot on the second generation devices.
Workaround: Modify the MEP configuration and the alarm and status will come back in sync.

Layer-2 loopback on second generation devices do not support IP SWAP
Layer-2 loopback on second generation devices do not support IP SWAP.

LBM gap counted but not packet loss
The LBM test reports gap count but the packet loss count is not supported.

Ant Advance Feature raises alarm for the unused media of the combo port
The Ant Advance Feature raises alarm for the unused media being down on the combo port. The alarm from the unused media should be suppressed.

JDSU loopback stays up after Loop-Up timer expiration
The JDSU loopback stays up after Loop-Up timer expiration in the second generation devices when the module management is down. When the module management is up, the timer expiration is working appropriately.

Drop opposite timeout not working on second generation devices
The loopback drop opposite timeout is not working on the second generation devices. The drop opposite traffic stays configured until the loopback is disabled or deleted.

Y.1564 performance test returns IR lower than CIR for test duration longer than eight (8) hours
Y.1564 performance test should be limited to a duration that is smaller than eight hours. Duration longer than eight hours could return IR lower than CIR.

Anto10G: Policy are still applied with TGEN feature
The policy configurations are still applied on Ant10G even when the active feature is TGEN.

The active feature does not switch automatically from TGEN to PMON when configuring DMM reflection
The active feature does not switch automatically from TGEN to PMON when configuring DMM reflection.
Workaround: Change the active feature manually or set the remote device to in-service with PMON as an active feature.

Port show status for auto-nego returns empty value
On Ant10G, when using the SFP speed override, the auto-nego field returns an empty value. The auto-negotiation is disabled in that case.

RFC-2544 generator is stopped completed but the flow state is running
The RFC-2544 generator flow state is running but the test is stopped and the stop time is reported.
Workaround: Stop the RFC-2544 test in order to restart it.

SFP edit force-link-up can be used on Ant Advanced Feature and Ant10G
The SFP edit command for force-link-up can be used in Ant Advanced Feature and Ant10G even if it is not supported.

SkyLIGHT VCX port name corrupted after upgrading from release VCX 2.1
Port names have a maximum length of 31 characters each. Assigning a long name to a port is problematic since the port name is a concatenation of the port name and “-intf0” or “intf1”. When the concatenated name is bigger than 31 characters, the name is corrupted. The maximum port name length should be 25 characters.

Nano2 with link partner in auto negotiation disable is reported with auto negotiation enable
The link partner auto negotiation state for Nano2 reports auto negotiation enable even if it is disabled on the link partner.

Link partner pause advertisement is not reported
The link partner pause advertisement is not reported.

Port status report pause capabilities even is not supported
The port status reports pause capabilities being supported even if the pause is not supported.

FlowBROKER Packet Engine statistics does not count all matching traffic at line rate
When the traffic being captured by FlowBROKER is at line rate, the captured traffic cannot be transmitted. The captured traffic has a lower priority than customer traffic. The captured packet count added to the missed packet count does not equal the on wire number of packets.

Monitoring Y.1564 test execution block port statistics and link status update
When monitoring Y.1564 test execution live using CLI or web interface, the port statistics and port link status of the remote device executing the Y.1564 test are not refreshed.
Workaround: Stop monitoring Y.1564 test for 10 seconds and port statistics and link status will be refreshed.

Rate adaptation issues with the FlowBROKER on ant Module (Advanced Features)
The FlowBROKER Packet Engine rejects the configuration on the ant Module (Advanced Features) when performing rate adaptation between UNI and NNI ports:

• The FlowBROKER Packet Engine can be configured when the NNI port link speed is slower than the UNI link speed.
• When the NNI port link speed is faster than or equal to the UNI port link speed, the FlowBROKER is functioning normally.

SD Manager Sync Master session type limitation
The NFV-based Sync Master cannot provide synchronization information to all session types; it can only synchronize sessions of the same type:

• The NFV TWAMP Sync Master can only synchronize NFV TWAMP sessions
• The NFV DMM Sync Master can only synchronize NFV DMM sessions

RFC-2544 burst mode with duration in seconds generates for a longer period
The burst generation period can be longer than expected when using RFC-2544 in burst mode if the duration has been specified in seconds.
Workaround: Express the preferred duration in number of packets.

Deleting the previous firmware suite on a device before the upgrade is completed
When upgrading a remote device, it is possible to delete the older firmware suite, even if the upgrade is still underway. It is strongly recommended to wait for the upgrade to complete before deleting the previous firmware suite on the remote device.

ARP requests from a VCE use the SkyLIGHT VCX’s management IP address
ARP requests from a VCE that does not have an assigned IP address uses the SkyLIGHT VCX’s management IP address as the source address. In such circumstances, no ARP requests should be sent.

NMP port statistics only updated every 60 seconds
Port statistics are only updated every 60 seconds when using SNMP. The SNMP statistics cache is updated every 60 seconds. The other statistics available via the Management Web Interface or the CLI are refreshed more frequently.

Reconfiguring an interface’s VLAN Priority stops its sessions in SkyLIGHT Director Manager
When the configuration of an interface is modified, the interface is deleted then recreated, thereby causing its sessions to stop. As such, reconfiguring an interface’s VLAN priority will cause any sessions using this interface to be stopped in SD Manager.
Workaround: Restart the affected sessions manually.

SkyLIGHT VCX TWAMP reflection should not use the remote device TWAMP reflection UDP port
When using SkyLIGHT VCX TWAMP software reflection, the TWAMP reflection UDP port must be different from the remote device TWAMP reflection UDP port for remote devices using NFV. If both have the same UDP port, both will reflect the TWAMP packet and cause packet duplication.

Host not reporting correct link speed for Nano Copper Advanced Feature when link speed change by the media link partner
The host does not always report the correct link speed for Nano copper Advanced Feature when the link speed is changed by the media link partner. The link speed stays at the previous link speed value. The link speed on the Nano copper Advanced Feature reports the correct speed.

ARP leak when VCE IP address is the same as remote device IP address
When the VCE IP address is the same as the remote device management IP address. Both interfaces respond to the ARP request if one of the interfaces has no VLAN tag. When both interfaces use different VLAN tag, there is no ARP leak.

SAT test cannot be started when updating the remote device
When the remote device is updating, the SAT test cannot be started. Once the update is completed, the SAT does not start and the test cannot be deleted.
Workaround: Restart SkyLIGHT VCX in order to start the SAT test.

Remote device management loss when remote device management and VCE use the same IP address and VLAN
The remote device management is lost when the management and VCE use the same IP address and VLAN. The remote device management IP address cannot be shared with VCE.

Unable to start multiple TWAMP sessions using the same source and destination IP address on different VCE
It is not possible to start multiple TWAMP sessions using the same source and destination IP address on different VCE.

VCE IP domain isolation route visible in CLI show active route
The VCE IP domain isolation routes are visible in the CLI show active route “VCE name”. These routes should be hidden.

Remote device names longer than 27 characters cause the port name to be truncated
When the remote device name exceeds 27 characters, the port name is truncated. The port name is limited to 32 characters; it contains the device name and appends either –UNI
or –NNI.
Workaround: The remote device names should be limited to 26 characters.

Management lost upon inserting media in the device’s preferred connector
Management is lost upon inserting media in the preferred connector of the ant Module (Advanced Features). When inserting the media in the preferred connector, the connector selection switches automatically to this connector. Management will be lost if the module cannot be managed using the preferred connector.

Error message when applying three-second discovery rate with the default timeout period
When applying a three-second discovery rate, an error message will be generated unless the default timeout is updated. The timeout rate must be shorter than or equal to the discovery rate.

Remote device management lost when rolling back to software without AMK support
The Accedian Management Key (AMK) ensures a universal authentication mechanism on Accedian Performance Modules; its value can be used to connect to all newly-shipped devices that have never been managed by a SkyLIGHT VCX.

When managing the remote device using the AMK as the security key, the remote device’s firmware cannot be rolled back to an older version unless it supports AMK. Attempting to roll back to a version that does not support AMK will cause the remote management to be lost. If this occurs, remote device management can be recovered using the per-unit security key (PUK) specific to a Performance Module.

Upgrading 1500 remote devices at the same time can take more than an hour
Upgrading 1500 remote devices at the same time can take up to 2.5 hours to complete. The upgrade process limits the number of parallel remote devices to 250 in order to avoid potential overload.

Cannot release license when in the clock windback grace period
A license cannot be released when in the clock windback grace period, since the license server goes offline whenever a clock windback is detected. The license can only be released after the grace period is over or when the clock windback has been resolved.

Frames with CRC error generated from the Modules
The Modules are designed by default to pass traffic in cut-through mode in order to reduce latency. In some cases, when Module management traffic is generated from the Modules towards an instance of the SkyLIGHT VCX, the traffic buffer can overflow and the next frame is transmitted in error. These frames are counted in the transmit CRC errored frames.

Loss of management when remote device’s interface is changed while managed at Layer 2
Changing the discovery instance going through an interface that is used to reach a remote device causes the interface to be deleted, then subsequently recreated. If a message intended for a Remote Device is sent while the interface is deleted, it will automatically set the remote device’s linked state to No. After 45 seconds the linked state should return to Initialized, then to Yes.

Remote device sends CCM frames with delay when linking to the SkyLIGHT VCX
Remote devices periodically send Continuity Check Messages (CCM) with delay when linking to the SkyLIGHT VCX. A delay of up to eight seconds can be observed. Once the remote device has been linked, the CCM frames are transmitted without further delay.

Degraded performance on some local ports with high ERSPAN throughput
When streaming FlowBROKER ERSPAN-captured traffic at 400 Mbps; frames on a 10G or 1G port, other than the Management port, may be lost.

FlowBROKER drops frames when 10 ERSPAN Type-3 are used as the Analyzer
When 10 or more ERSPAN Type-3 Analyzers are configured in an instance of the SkyLIGHT VCX, the FlowBROKER solution may lose some captured frames.

FlowBROKER ERSPAN Analyzer drops packets when FSX streams burst packets at line rate
Although FSX Performance Elements can stream captured packets at line rate in a burst, doing so may result in the FlowBROKER ERSPAN Analyzer dropping frames in the SkyLIGHT VCX.

FlowBROKER ERSPAN Analyzer drops packets in heavy traffic conditions
Packets may be dropped by the FlowBROKER ERSPAN Analyzer if the total captured bandwidth exceeds 200 Mbps.

Unable to disable the SNMP trap receiver
The SNMP trap receiver cannot be disabled; the trap is sent regardless of the Enable state.
Workaround: Remove the trap notification receiver’s IP address from the trap receiver configuration.

Packets missing from FlowBROKER ERSPAN Type-3 traffic
ERSPAN-streamed flows may be missing some captured packets.

Change of remote device management VLAN not protected by rollback
If a remote device is not reachable once its management VLAN has been modified, the change should be reverted in order to avoid losing management with the remote device. The rollback mechanism to revert the VLAN change has not yet been implemented.

Disabling a policy can cause a packet to be dropped
When disabling a policy, if a packet matches a field present in that policy, the packet can be dropped instead of passing through.

Remote Device Table update using SNMP can take up to 30 seconds
When the remote device table is updated, the update can take up to 30 seconds to appear in an SNMP client. The CLI and Web interfaces do not present this issue.

File size limitation when importing remote devices via Web interface
Importing remote devices via the Web GUI Interface does not function as expected when the file containing the remote device details is less than 512 bytes in size. To import small-sized remote device files, use the CLI import commands instead.

Cannot delete MEP if the connection to the device is lost
An active MEP cannot be deleted if the connection to its Performance Module has been lost. The Module must be detected and its connection valid (i.e., its Linked state must be Yes) before the MEP can be deleted.

Auto interface is unavailable in the DHCP interface selection
The Auto interface cannot be selected as the DHCP interface, or used for DHCP.

NTP alarms remain active after disabling NTP or deleting the NTP server entry
NTP alarms remain active, even after disabling the NTP functionality or deleting the NTP server used for synchronization.

NTP Client Lost Server alarm message does not refresh the server IP address
The NTP Client Lost Server alarm message is not refreshed when the NTP server is switched. The alarm message retains the IP address of the original NTP server.

Remote Device Import not working using TFTP
The remote device CSV file import does not function as expected using the TFTP file transfer protocol.
Workaround: Use the FTP, HTTP, SFTP or FTPS file transfer protocols instead.

Layer-3 traffic loopback by TWAMP reflector
RFC-2544 Layer-3 traffic can be looped back by the TWAMP reflector if both use the same UDP port. RFC-2544 traffic should therefore be generated on a different port than the TWAMP reflection port.