Known Issues
This section lists the known issues in this version of Mellanox NEO ® with available workarounds.
|
Ref. # |
Issue |
|
2388786 |
Description: Network map periodically refreshes. When that happens, selected items become unselected. |
|
Workaround: N/A |
|
|
Keywords: Network Map |
|
|
Detected in version: 2.7 |
|
|
2412055 |
Description: The What Just Happened button "Export to CSV" is missing from UI. |
|
Workaround: N/A |
|
|
Keywords: WJH, WebUI |
|
|
Detected in version: 2.7 |
|
|
2187852 |
Description: Any change related to the "RoCE" sub-category under "Events Policy" on NEO 2.4 or below will be lost upon NEO upgrade. User is required to reconfigure it after the upgrade. |
|
Workaround: N/A |
|
|
Keywords: RoCE, upgrade |
|
|
Detected in version: 2.7 |
|
|
2107670 |
Description: When working with NEO, any switch configuration done directly on the switch (via switch CLI) might be conflicted with NEO configuration and interfere with NEO switch management and configuration. |
|
Workaround: N/A |
|
|
Keywords: NEO, CLI, configuration, conflict, overwrite |
|
|
Detected in version: 2.7 |
|
|
2225134 |
Description: In Streaming Settings, the events "TTL value is too small" and "Packet size is larger than MTU" (under Forwarding > L3) are always streamed even if configured otherwise. |
|
Workaround: N/A |
|
|
Keywords: Streaming, MTU, TTL |
|
|
Detected in version: 2.7 |
|
|
2147008 |
Description: RoCE service does not display MLAG port-channel traffic. |
|
Workaround: N/A |
|
|
Keywords: RoCE, MPo |
|
|
Detected in version: 2.7 |
|
|
2211606 |
Description: The NEO dashboard, the WJH dashboard specifically, may at times become slow to respond. |
|
Workaround: N/A |
|
|
Keywords: WJH, dashboard, slow |
|
|
Detected in version: 2.7 |
|
|
2248362 |
Description: Upgrading Mellanox NEO to software version 2.7 rebuilds the database and removes all Telemetry and WJH data. |
|
Workaround: N/A |
|
|
Keywords: WJH, database |
|
|
Detected in version: 2.7 |
|
|
2239514 |
Description: Any task with a snapshot created on top of it in NEO 2.4 version (or older) is displayed under "Telemetry" → "Snapshots", not under "Tasks". |
|
Workaround: N/A |
|
|
Keywords: Telemetry, task, snapshot |
|
|
Detected in version: 2.7 |
|
|
2239662 |
Description: Any collector added to a telemetry session in NEO version 2.4 (or older) is detached from the session after Mellanox NEO upgrade. |
|
Workaround: Reattach collector after software upgrade. |
|
|
Keywords: Collector, session, upgrade |
|
|
Detected in version: 2.7 |
|
|
2245329 |
Description: WJH buffer drop trap_probability is probability to extract the packet from stream of the packets that get exception. If the packet rate is low the extraction rate can deviate from configured value. |
|
Workaround: N/A |
|
|
Keywords: WJH |
|
|
Detected in version: 2.7 |
|
|
2098905 |
Description: NEO OpenStack integration is supported only over HTTPS protocol (HTTP is not supported). |
|
Workaround: N/A |
|
|
Keywords: HTTPS, OpenStack |
|
|
Detected in version: 2.6 |
|
|
2119441 |
Description: NEO supports up to to 10 MPOs in a single "apply MLAG" service. |
|
Workaround: If more MPOs are required, the service must be updated and re-applied. |
|
|
Keywords: MLAG, MPO |
|
|
Detected in version: 2.6 |
|
|
- |
Description: When interoperating with switch systems installed with Onyx 3.9.0300, they may reach high CPU utilization. |
|
Workaround: Run the command "ssh server login record-period 1" in order to avoid this. |
|
|
Keywords: Onyx, high CPU utilization |
|
|
Detected in version: 2.6 |
|
|
2126093 |
Description: Mellanox Onyx ® switches support up to 64 buffer histogram samplings. |
|
Workaround: N/A |
|
|
Keywords: Histogram, buffer events, telemetry |
|
|
Detected in version: 2.6 |
|
|
2118673 |
Description: The following 3rd party systems are not supported by Mellanox NEO ® : Arista, Brocade, Cisco, Juniper and HP. It is not possible to add new systems from these vendors to NEO. However, if NEO is upgraded from an older version where these switch systems have been added, then they can be presented and managed without issue. NEO is still able to detect these 3rd party systems by IP range scanning or by LLDP. |
|
Workaround: N/A |
|
|
Keywords: 3rd party, switch systems |
|
|
Detected in version: 2.6 |
|
|
2107670 |
Description: Switch configuration performed directly on the switches (via switch CLI) may conflict with Mellanox NEO configuration and interfere with NEO switch management and configuration. |
|
Workaround: Only use NEO for managing configuration over the managed switches (avoid manual configuration on the managed switches). |
|
|
Keywords: Configuration, CLI |
|
|
Detected in version: 2.6 |
|
|
1918927 |
Description: In-band migration not supported. |
|
Workaround: N/A |
|
|
Keywords: VLAN provisioning |
|
|
Detected in version: 2.6 |
|
|
1912682 |
Description: Telemetry Agent does not provide telemetry information on split ports if they are configured while the agent is running. |
|
Workaround: Restart Telemetry Agent. |
|
|
Keywords: Telemetry Agent, split ports |
|
|
Detected in version: 2.6 |
|
|
1952279 |
Description: Every new device added to NEO must have a unique management IPv4 address, otherwise the displayed devices data might be corrupted. |
|
Workaround: N/A |
|
|
Keywords: Management Elements |
|
|
Detected in version: 2.6 |
|
|
2082427 |
Description: When configuring LLDP on the host interface, LLDP must be configured to publish the host management IPv4 address. Otherwise, the host is not presented correctly in Mellanox NEO. |
|
Workaround: N/A |
|
|
Keywords: LLDP |
|
|
Detected in version: 2.6 |
|
|
2061726 |
Description: When adding a Cumulus switch to Mellanox NEO, the initial discovery results in SNMP failure. Once SNMP is configured on the switch, it returns to normal status (i.e. "OK"). |
|
Workaround: N/A |
|
|
Keywords: Cumulus, SNMP |
|
|
Detected in version: 2.6 |
|
|
– |
Description: NEO v2.5.1 supports up to 50 managed switches. If Mellanox NEO is managing more than 20 switches, it is recommended to use SSD disk otherwise NEO performance issues are expected. |
|
Workaround: N/A |
|
|
Keywords: Managed switches, performance |
|
|
Detected in version: 2.5.1 |
|
|
– |
Description: Running the Telemetry Agent configuration provisioning templates will attempt to restart the Agent, but it will fail to start it. Affected Telemetry Agent Provisioning Templates:
|
|
Workaround: Edit the template in NEO:
|
|
|
Keywords: Telemetry Agent Provisioning Templates |
|
|
Detected in version: 2.5.1 |
|
|
– |
Description: Running WJH on a Cumulus switch is not supported. (Up to the release of NEO v2.5, no Cumulus version that supports WJH existed.) |
|
Workaround: N/A |
|
|
Keywords: WHJ, Support, Cumulus |
|
|
Detected in version: 2.5 |
|
|
– |
Description: Running WJH and Threshold Events telemetry sessions on Onyx switches is supported only for Onyx version 3.8.2004 or newer. |
|
Workaround: N/A |
|
|
Keywords: WJH, Threshold events, Onyx Version |
|
|
Detected in version: 2.5 |
|
|
1922607 |
Description: If a device (Linux host or switch) is removed from NEO while some Mellanox switches are running telemetry, then all the telemetry sessions running on these switches will be stopped. |
|
Workaround: Manually disable and enable telemetry sessions using NEO (Telemetry → Streaming) in order to reactivate the required telemetry sessions. |
|
|
Keywords: Telemetry, Device, Remove, Switch, Session, Stop |
|
|
Detected in version: 2.5 |
|
|
1917681 |
Description: NEO monitoring over SNMP is not supported for Cumulus switches (due to a known issue in Cumulus switch). |
|
Workaround: N/A |
|
|
Keywords: SNMP, Monitoring, Cumulus |
|
|
Detected in version: 2.5 |
|
|
1917323 |
Description: If a switch is unresponsive, NEO will not display a continuous graph of the monitoring data. |
|
Workaround: N/A |
|
|
Keywords: Unresponsive, Switch, Continuous, Monitoring |
|
|
Detected in version: 2.5 |
|
|
1920182 |
Description: General device information (Memory and CPU) might be displayed at a delay of 2-4 minutes after the device has been added to NEO. |
|
Workaround: N/A |
|
|
Keywords: General information, Memory, CPU, Delay |
|
|
Detected in version: 2.5 |
|
|
1920520 |
Description: In Cumulus switch, in case of “non-ascii” characters used in the switch configuration files, creating configuration backup and network snapshots or restoring from them might fail. |
|
Workaround: N/A |
|
|
Keywords: Cumulus, “non-ascii”, Characters, Configuration |
|
|
Detected in version: 2.5 |
|
|
1920601 |
Description: When editing MLAG port channels via MLAG wizard, configuration changes might fail in case telemetry was configured prior to the change (e.g., via the Bring-Up Wizard). |
|
Workaround: N/A |
|
|
Keywords: MLAG, Editing, Configuration, Telemetry |
|
|
Detected in version: 2.5 |
|
|
1887761 |
Description: Telemetry agent will not publish telemetry data for MLAG port channel in the following cases:
In these cases, telemetry data is published for the physical ports (the MLAG port channel members). |
|
Workaround: N/A |
|
|
Keywords: MLAG, Telemetry, MLAG port channel, disabled, enabled |
|
|
Detected in version: 2.5 |
|
|
– |
Description: When upgrading NEO v2.4 to NEO v2.5, due to the transition from Graphite to InfluxDB, historical counters data kept on Graphite will not be transferred to the InfluxDB. |
|
Workaround: N/A |
|
|
Keywords: Upgrade, Graphite, InfluxDB, counters |
|
|
Detected in version: 2.5 |
|
|
1848870 |
Description: General information (CPU and Memory information) for Cumulus switches managed by NEO are not displayed in the NEO interface until it is exposed by the switch. For more information, please refer to Exposing CPU and Memory Information via SNMP. |
|
Workaround: N/A |
|
|
Keywords: CPU and Memory information, Cumulus switches |
|
|
Detected in version: 2.5 |
|
|
– |
Description: RoCE Service configuration is not supported for Onyx versions prior to 3.6.5000. |
|
Workaround: Upgrade the switch to the latest Onyx version. |
|
|
Keywords: Services, RoCE, Onyx |
|
|
Detected in version: 2.5 |
|
|
– |
Description: RoCE Service configuration cleanup is not supported for services upgraded from older NEO versions. |
|
Workaround: Remove the old service and recreate it with the latest NEO. |
|
|
Keywords: Services, Clean-up, RoCE, upgrade |
|
|
Detected in version: 2.5 |
|
|
– |
Description: RoCE Service configuration cleanup is supported only for Onyx and Cumulus switches. |
|
Workaround: For other service types, remove the configuration manually using switch CLI. |
|
|
Keywords: Services, Clean-up, RoCE |
|
|
Detected in version: 2.5 |
|
|
– |
Description: Service configuration clean-up is supported only for RoCE service. |
|
Workaround: For other service types, remove the configuration manually using switch CLI. |
|
|
Keywords: Services, Clean-up |
|
|
Detected in version: 2.5 |
|
|
1329530 |
Description: Manual HA takeover or failover might take up to 60 seconds (depending on the machine NEO is running on). During that time, triggering additional failover or takeover operations might result in the original action failure. |
|
Workaround: Wait for at least 60 seconds between HA operations – failover or takeover. |
|
|
Keywords: HA, failover, takeover |
|
|
Detected in version: 2.4 |
|
|
1600868 |
Description: Configuring RoCE on a host bond interface is currently not supported. |
|
Workaround: Configure RoCE on the bond slaves. |
|
|
Keywords: Bond, RoCE |
|
|
Detected in version: 2.3 |
|
|
1582800 |
Description: Running too many frequent live monitoring sessions for a specific switch may overload the switch's JSON API and result in timeouts. |
|
Workaround: Run fewer live monitoring sessions in parallel. |
|
|
Keywords: JSON, timeout, live monitoring |
|
|
Detected in version: 2.3 |
|
|
– |
Description: When a WJH is enabled on the Telemetry Agent, WJH on the Onyx switch is disabled (the user is not able to view WJH details via Onyx switch CLI) and vice versa. |
|
Workaround: N/A |
|
|
Keywords: What Just Happened, WJH, Onyx, Telemetry Agent |
|
|
Detected in version: 2.3 |
|
|
2090123 |
Description: WJH is supported by NEO only for Onyx Spectrum switches using v3.7.1134, or newer. |
|
Workaround: N/A |
|
|
Keywords: What Just Happened, WJH, Dropped Packets |
|
|
Detected in version: 2.3 |
|
|
– |
Description: NEO-Host installation is supported only for Linux hosts, using one of the following HCAs: ConnectX-4 / ConnectX-4 Lx / ConnectX-5. |
|
Workaround: N/A |
|
|
Keywords: NEO-Host |
|
|
Detected in Version: 2.3 |
|
|
1578231 |
Description: NEO telemetry agent can stream Routing Table information up to 20K records, and MAC table information up to 800 records. |
|
Workaround: N/A |
|
|
Keywords: Telemetry Agent, Routing Table, MAC Table |
|
|
Detected in Version: 2.3 |
|
|
1417273 |
Description: System icons are not shown for Edge and Safari systems. |
|
Workaround: N/A |
|
|
Keywords: Network Map, Edge, Safari |
|
|
Detected in Version: 2.2 |
|
|
1504128 |
Description: The network path calculation requires that all switches along the path will have the same SSH credentials. Otherwise, the calculation will fail. |
|
Workaround: N/A |
|
|
Keywords: Network Path, SSH, Credentials |
|
|
1484291 |
Description: The telemetry agent cannot be stopped on switches running Onyx OS v3.6.8100. |
|
Workaround: Do not deploy the telemetry agent on Onyx OS v3.6.8100. |
|
|
Keywords: Telemetry Agent, Onyx |
|
|
1421369 |
Description: The "In Packets rate" calculated counter shows an incorrect value for Cumulus switches only, due to an issue with the switch (the Unicast RX Packets counter always returns a value of zero). |
|
Workaround: N/A |
|
|
Keywords: In Packets Rate, Cumulus, Unicast RX Packets |
|
|
1498434 |
Description: The network path calculation will display the links transmitted bandwidth utilization according to the maximal value of the aggregated links (in case of a multiple links connection). |
|
Workaround: N/A |
|
|
Keywords: Network Path, Bandwidth, Utilization |
|
|
1332120 |
Description: Telemetry Agent does not support split ports. |
|
Workaround: N/A |
|
|
Keywords: Telemetry Agent, Split Port |
|
|
1328501 |
Description: In the MLAG service, the bond is configured with the default gateway. |
|
Workaround: Configure a different static route to the relevant ports. |
|
|
Keywords: MLAG, Bond |
|
|
1316429 |
Description: Port live monitoring only works from a certain Onyx version. |
|
Workaround: N/A |
|
|
Keywords: Telemetry, Live Monitoring |
|
|
1327385 |
Description: Upgrade procedure (from an older version to 2.1.0) does not include Events Policy and RoCE Service. |
|
Workaround: N/A |
|
|
Keywords: Upgrade |
|
|
1309655 |
Description: Telemetry session interval cannot be changed |
|
Workaround: N/A |
|
|
Keywords: Telemetry Agent |
|
|
1298137 |
Description: When loading images with a similar name (differed only by tag) the 1st image name becomes empty due to an issue in Red Hat Docker. |
|
Workaround: N/A |
|
|
Keywords: Docker, Container |
|
|
1272497 |
Description: There is no validation for the maximum ECN value in RoCE Service. The max allowed ECN value is dynamic and depends on switch type, current memory state , etc. |
|
Workaround: N/A |
|
|
Keywords: RoCE Service |
|
|
1277047 |
Description: Configuring one of the IPL ports in MLAG service to 'switchport mode trunk' fails the service. |
|
Workaround: Reset switchport mode before adding the port to the IPL. |
|
|
Keywords: MLAG Service |
|
|
1302777 |
Description: Switch reboot stops a telemetry agent session (if running). |
|
Workaround: After switch reboot, manually restart the telemetry session. |
|
|
Keywords: Telemetry |
|
|
1071652 |
Description: For optimized UI functionalities, LastPass browser add-on should either be disabled or not installed. |
|
Workaround: N/A |
|
|
Keywords: UI, LastPass |
|
|
– |
Description: A NEO-Host package installation is required for successful provisioning of RoCE through the new RoCE service. |
|
Workaround: Install NEO-HOST either on Linux-without-Neo-Host-installed predefined group or on a specific host. |
|
|
Keywords: RoCE Service |
|
|
1064979 |
Description: The MLAG service is supported in MLNX Onyx (MLNX_OS) starting from v3.6.4000. |
|
Workaround: Make sure to upgrade your Onyx version to v3.6.4000 or above. |
|
|
Keywords: MLAG Service |
|
|
– |
Description: When using SNMPv3 with sha authentication and priv=aes128 option, the switch will become unreachable due to timeout. |
|
Workaround: For Mellanox PPC switches, use md5 authentication with a priv=des option. |
|
|
Keywords: Authentication |
|
|
– |
Description: Mellanox NEO Client (browser) might fail to connect to the NEO server in case the iptables service is running. |
|
Workaround: Make sure to disable the iptables service before running NEO installation. |
|
|
Keywords: Installation |
|
|
– |
Description: VLANs and LAGs information may not be displayed as part of device information for non-Mellanox devices. |
|
Workaround: N/A |
|
|
Keywords: 3rd Party Systems Support |
|
|
– |
Description: Linux/Windows host provisioning via NEO is non-persistent. |
|
Workaround: N/A |
|
|
Keywords: Host Provisioning |
|
|
– |
Description: NEO start-up will fail in case the machines’ local time zone is not configured. |
|
Workaround: Make sure the installed machines’ local time zone is configured. (/etc/localtime file exists). |
|
|
Keywords: NEO Start-Up |
|
|
– |
Description: Apply Config operation is only available for switches with Onyx v3.6.2000 and above. |
|
Workaround: N/A |
|
|
Keywords: Configuration Management |
|
|
– |
Description: Cable information is only supported for Mellanox Onyx switch ports. |
|
Workaround: N/A |
|
|
Keywords: Cable Information |
|
|
951789 |
Description: Performance tests are only supported for ConnectX-4 and ConnectX-5 family adapter cards. |
|
Workaround: N/A |
|
|
Keywords: Performance Check |
|
|
– |
Description: Performance check can be performed only on two Linux hosts, running MLNX_OFED_LINUX-3.3-1.0.4.0 version or higher. |
|
Workaround: N/A |
|
|
Keywords: Performance Check |
|
|
– |
Description: RoCE configuration on hosts is non-persistent. |
|
Workaround: N/A |
|
|
Keywords: RoCE Service |