A-Series DAC Cable issue

[aali]

One of our customers recently puchased (2) HP A5900 Series switches with A-Series specific DAC 1.2M DAC cables.  They are experiencing a high-rate packet loss and am just wondering if the issue is with the DAC or configuration.

See below the issues that customer is reporting.

We have connected servers (VMware and Linux) that have high packet losses from ping. Sometimes this is fixed by reseating cables sometimes not. I currently have a Linux machine that isn’t working due to high packet loss

We have a VMware host that shows slow network performance – VMotion takes 20+ minutes for a workload that finishes in less than a minute on the other VMware hosts on the 10G switch.

We have one cable/switch port that hangs the network on a VMware host if it is plugged in. We assume that something is wrong with the cable, but we don’t understand why one bad interface would bring down VMware networking.

All the hosts are DL360 Gen 8 servers with HP DAC cables

Can someone shed a light?

[LorenzoCastro]

Hello, Are these the only hosts pluged into the switch and having the problems? Doesn't sound like it has anything to do with the cables specifically, but rather a L2 loop that is causing the issues.  Have you reviewed any logs on your switches? How are the switches connected to each other in an IRF stack or just uplinked to each other via the DACs?   If uplinked, do you have STP enabled?

[Richard Litchfield]

Is IRF configured on the 5900 pair so that it acts as a single virtual switch? (they may have a 40Gb QSFP+ DAC that connects them together).

If they are not running IRF, change config to include IRF to maximise the benefits of these switches - eg aggregration from a single server to both switches, without worrying about loops.

[aali]

Thanks folks.  When I left the site, the IRF was functional, but need to go back and verify if someone didn't make any changes.  yes, I didn't think of that but I belive as well that there is a loop in the network.

Regards,