You are experiencing intermittent connectivity issues on one specific Cisco UCS server Some applications work, while other do not transfer data You suspect jumbo frames limitations Which two places do

Question

Options

• ACheck server's service profile and/or service profile template
• BCheck server's QoS System Class
• CCheck the fibre channel port configuration for jumbo frame configuration
• DCheck upstream switch ports frame size configuration
• ECheck server's individual vNICs and/or vHBAs properties

• A
  3% (1)
• B
  74% (25)
• C
  9% (3)
• E
  15% (5)

Explanation

Jumbo frame MTU configuration in Cisco UCS is controlled at the QoS System Class level (B), not at individual vNIC/service profile level - this is the centralized place in UCS Manager where you set MTU (typically 9000 bytes) for Ethernet classes of service. Because traffic must traverse the entire path end-to-end, the upstream switch ports (D) must also be configured to accept oversized frames; if the Nexus or other connected switch drops frames above 1500 bytes, intermittent application failures occur (larger-payload apps fail, smaller-payload apps succeed - exactly the symptom described).

A (service profile/template) is wrong because jumbo frames aren't configured there directly; the service profile references QoS policies but doesn't set MTU itself. C (Fibre Channel port config) is wrong because jumbo frames are an Ethernet/IP concept - FC uses its own framing and is unrelated to this symptom. E (individual vNICs/vHBAs) is wrong because Cisco UCS does not expose per-vNIC MTU settings; MTU is inherited from the QoS System Class, making E a common trap answer.

Memory tip: Think "Q-U-E-U-E" - QoS (UCS side) + Upstream switch = the two-hop queue jumbo frames must clear. If either hop rejects the large frame, your high-bandwidth apps break while lightweight ones sail through.

No community discussion yet for this question.