Hi!
Having a very strange behaviour between six HP DL380p G9 servers and a Cisco C4900M switch
connected with DAC-cables (Cisco H10GB-CU1M), but also tested connection with a fibre cable with Cisco-coded SFP:s (SFP+) and got same result.
Problem: The C4900M reporting “err-disable” on some of the ports (not all) and a high packet loss (for testing ping between all servers). The “err-disable” cause is “Link flaps”. Some of the servers is always giving “err-dissable”, some of them doing sometimes and some of them is never giving that error. Vmkernal log reporting the nic-card is going up an down.
Troubleshooting: The following tests is done, followed by the result.
- Checked that all setup, config and versions of firmware, software is the same for the six HP servers and the six ports on the switch.
- Connecting two servers directly to each other through DAC-cable. Result: no problems at all (0% packet loss), and ports is up all time.
- Cross connect two ports on switch with DAC-cable (tested several cables and ports). Result: no “Link flaps” reported. Haven´t tested to ping.
- Connected servers to switch with DAC-cable. Result: Some (of the six) ports on switch is reporting “Link flaps”, between 5-60% packet loss during ping from different servers and from switch.
- Moved cable from one connection that not reporting “Link flaps” to one that is reporting. Result: No change (same server still reporting “Link flaps”).
- Switched port on switch between one server that is reporting “Link flaps” and one that not is doing. Result: Same server still reporting “Link flaps”.
- Switched CVR-X2-SFP10G adaptors (X2->SFP+) between to switch port that is reporting and not reporting “Link Flaps”. Result: No change.
Switch: A standalone C4900M with no other connections than the six servers (ok, management port is connected to a LAN).
IOS-version: c4500e-entservicesk9-122.53.SG3. Ports settings is:
Port config: switchport mode access, switchport access vlan 1
Servers: All new HP DL380p G9 with ESXI 6.5 installed. Using one of the two 10 GB SFP+ ports of the “built in” 560FLR-card.
HP System ROM: P89 v2.56 (01/22/2018)
HPE Ethernet 10GB 2-port 560FLR-SFP+Adapter (driver: ixgbe; version: 4.5.2-iov; firmware-version: 0x80000838, 1.1752.0)
VMware ESXi 6.5.0 build-7967591, VMware ESXi 6.5.0 Update 1
Have anyone an idea of root cause for the issue?