NVIDIA UFM Enterprise User Manual v6.17.0
NVIDIA UFM Enterprise User Manual v6.17.0

Subnet Manager Tab

UFM is a management platform using a user-space application for InfiniBand fabric management. This application is developed within the context of an open-source environment. This application serves as an InfiniBand Subnet Manager and a Subnet Administration tool.

The UFM Subnet Manager (SM) is a centralized entity running on the server that discovers and configures all the InfiniBand fabric devices to enable traffic flow throughout the fabric.

To view and configure SM parameters in the Subnet Manager tab, select the relevant tab according to the required configuration.

For more information, please refer to Appendix – Enhanced Quality of Service.

The SM Keys tab enables you to view the Subnet Manager Keys. You cannot change the configuration in this tab.

image2019-6-16_15-0-0-version-1-modificationdate-1713267876717-api-v2.png

Field

Description

Default

MKey

A field that allows you to view or edit the M_Key value sent to all ports to qualify all the set (PortInfo). Authentication is performed by the management entity at the destination port and is achieved by comparing the key contained in the SMP with the key (the M_Key Management key) residing at the destination port.

0x0000000000000000

SA Key

Shows the SM_Key value to qualify the receive SA queries as 'trusted'.

0x0000000000000001

Subnet Prefix

An identifier of the subnet. The subnet prefix is used as the most significant 64 bit of the GID of each InfiniBand node in the subnet.

0xfe80000000000000

SM Key

Read-only field that displays the Key of the Subnet Manager (SM).

0x0000000000000001

MKey Lease Period

A field that allows you to view or edit the lease period used for the M_Key on this subnet in [sec].

0

LMC

Defines the LID Mask Control value for the SM. Possible values are 0 to 7.

LID Mask Control (LMC) allows you to assign more than one LID per port.

NOTE: Changes to the LMC parameter require a UFM restart.

0

No Partition Enforcement

Disables partition enforcement by switches.

Disabled

The SM Limits tab enables you to view and set the Subnet Manager Limits.

image2019-6-16_15-1-24-version-1-modificationdate-1713267878637-api-v2.png

To configure SM Limits, set the fields as described in the table below, and click “Save."

Field

Description

Default

Packet Life Time

A field that allows you to view and/or edit the code of maximum lifetime a packet in a switch.

The actual time is 4.096 usec * 2^<packet_life_time>.

The value 0x14 disables this mechanism

0x12

Subnet Timeout

A field that allows you to view and/or edit the subnet_timeout code that will be set for all the ports. The actual timeout is 4.096usec * 2^<subnet_timeout>

18

Maximal Operational VL

A field that allows you to view and/or edit the limit of the maximal operational VLs:

  • 0: NO_CHANGE

  • 1: VL0 1

  • 2: VL0_VL1

  • 3: VL0_VL3

  • 4: VL0_VL7

  • 5: VL0_VL14

3

Head of Queue Life Time

A field that allows you to view and/or edit the code of maximal time a packet can wait at the head of transmission queue. The actual time is 4.096usec * 2^<head of queue lifetime>

The value 0x14 disables this mechanism.

0x12

Leaf Head of Queue Life Time

A field that allows you to view and/or edit the maximum time a packet can wait at the head of queue on a switch port connected to a CA or gateway port.

0x10

VL Stall Count

A field that allows you to view the number of sequential packets dropped that cause the port to enter the VLStalled state. The result of setting this value to zero is undefined.

0x07

Leaf VL Stall Count

This field allows you to view the number of sequential packets dropped that cause the port to enter the VLStalled state. This value is for switch ports driving a CA or gateway port. The result of setting the parameter to zero is undefined.

0x07

Force Link Speed

A parameter that allows you to modify the PortInfo:LinkSpeedEnabled field on switch ports.

If 0, do not modify.

  • Values are:

  • 1: 2.5 Gbps

  • 3: 2.5 or 5.0 Gbps

  • 5: 2.5 or 10.0 Gbps

  • 7: 2.5 or 5.0 or 10.0 Gbps

  • 2,4,6,8-14 Reserved

  • 15: set to PortInfo:LinkSpeedSupported

15

By default, UFM sets the enabled link speed equal to the supported link speed.

Local Physical Error Threshold

A field that allows you to view and/or edit the threshold of local phy errors for sending Trap 129.

0x08

Overrun Errors Threshold

A field that allows you to view and/or edit the threshold of credit overrun errors for sending Trap 130.

0x08

Note

This tab is available to users with an advanced license only.

The SM Lossy tab enables you to view and set the Lossy Configuration Manager options after Lossy Configuration has been enabled.

image2019-6-16_15-3-3-version-1-modificationdate-1713267878307-api-v2.png

The SM SL2VL tab enables you to view the SL (service level) to VL (virtual lane) mappings and the configured Lossy Management. You cannot change the configuration in this tab.

However, you can change it in the previous SM Lossy Manager Configuration (Advanced License only) tab.

image2019-6-16_15-3-42-version-1-modificationdate-1713267879790-api-v2.png

The Sweep tab enables you to view and/or set the Subnet Manager Sweep Configuration parameters.

image2019-6-16_15-4-12-version-1-modificationdate-1713267882323-api-v2.png

To configure SM Sweep, set the fields as described in the table below and click “Save."

Field/Box

Description

Default

Sweep Interval

A field that allows you to view and/or edit the number of seconds between light sweeps (0 disables it).

10

Reassign LIDs

If enabled, causes all LIDs to be reassigned.

Disabled

Sweep on Trap

If enabled, traps 128 and 144 will cause a heavy sweep.

Enabled

Force Heavy Sweep

If enabled, forces every sweep to be a heavy sweep.

Disabled

The SM Handover tab enables you to view the Subnet Manager Handover Configuration parameters. You cannot change the configuration in this tab.

image2019-6-16_15-5-50-version-1-modificationdate-1713267880113-api-v2.png

Field/Box

Description

Default

SM Priority

A field that shows the SM priority used for determining the master. Range is 0 (lowest priority) to 15 (highest). Note: Currently, these settings may not be changed.

15

Polling Timeout

A field that shows the timeout in [sec] between two polls of active master SM.

Range=10000

Polling Retries

Number of failing polls of remote SM that declares it "not operational."

4

Honor GUID to LID File

If enabled, honor the guid2lid file when coming out of standby state, if the file exists and is valid.

Disabled

Ignore other SMs

If enabled, other SMs on the subnet are ignored.

Disabled

The SM Threading tab enables you to view the Subnet Manager Timing and Threading Configuration parameters. You cannot change the configuration in this tab.

image2019-6-16_15-7-9-version-1-modificationdate-1713267880437-api-v2.png

Field/Box

Description

Default

Max Wire SMPs

A field that shows the maximum number of SMPs sent in parallel.

4

Transaction Timeout

A field that shows the maximum time in [msec] allowed for a transaction to complete.

200

Max Message FIFO Timeout

A field that shows the maximum time in [msec] a message can stay in the incoming message queue.

10000

Single Thread

When enabled, a single thread is used for handling SA queries.

Disabled

The SM Logging tab enables you to view and/or set the Subnet Manager Logging Configuration parameters.

image2019-6-16_15-8-7-version-1-modificationdate-1713267882630-api-v2.png

To configure SM Logging, set the fields as described in the table below and click “Save."

Field/Box

Description

Default

Log File

Path of the Log file to be used.

cond/opt/ufm/files/log/opensm.log

Log Max Size

A field that allows you to view and/or edit the size limit of the log file in MB. If overrun, the log is restarted.

4096

Dump Files Directory

The directory that holds the SM dump file.

/opt/ufm/files/log

Force Log Flush

Force flush to the log file for each log message.

Disabled

Accumulate Log File

If enabled, the log accumulates over multiple SM sessions.

Enabled

Log Levels

Available log levels: Error, Info, Verbose, Debug, Funcs, Frames, Routing, and Sys.

Error and Info

The Misc tab enables you to view additional Subnet Manager Configuration parameters. You cannot change the configuration in this tab.

image2019-6-16_15-9-10-version-1-modificationdate-1713267880880-api-v2.png

Field/Box

Description

Default

Node Names Map File

A field that allows you to view and/or set the node name map for mapping nodes to more descriptive node descriptions.

None

SA Database File

SA database file name

None

No Clients Reregistration

If enabled, disables client re-registration.

Disabled

Disable Multicast

If enabled, the SM disables multicast support and no multicast routing is performed.

Disabled

Exit on Fatal Event

If enabled, the SM exits on fatal initialization issues.

Enabled

The QoS tab allows you to enable or disable QoS functionality. QoS is disabled by default.

QoS-version-1-modificationdate-1713267882000-api-v2.PNG

The Congestion Control tab allows you to enable, disable, or ignore congestion control.

image-2024-2-4_10-20-32-version-1-modificationdate-1713267875997-api-v2.png

The Adaptive Routing tab allows you to configure adaptive routing parameters.

Adaptive_Routing-version-1-modificationdate-1713267883503-api-v2.png

© Copyright 2024, NVIDIA. Last updated on Aug 27, 2024.