Skip to main content
Start a Conversation

Unsolved

This post is more than 5 years old

nickirwin

4 Posts

41370

July 23rd, 2014 08:00

Packet Loss on PowerEdge R720 with Broadcom 5720

Hi All, This is not your usual sg3/ESX/Windows driver issue...... We have 4 vSphere 5.5 Servers running on the same hardware (R720 with On-Board Broadcom 5720s) and during a reboot the entire switch its connected to gets flooded and we experience huge packet loss to anything that is connect to the switch. After the Lifecycle Controller initiates (just before it boots to the OS), its fine and it's also fine within vSphere - never drops a packet. We have updated all Lifecycle controllers/BIOS's to the latest firmware, but nothing seems to resolve it - it also happens if the server is physically powered down but has power. This happens on all 4 vSphere boxes and it cannot be a fault with all 4 servers - anyone have any idea? or seen this before? Thanks

Anonymous

5 Practitioner

 • 

274.2K Posts

July 23rd, 2014 13:00

What switch are you using? Is the iDRAC from the server plugged into the switch? How are the ports on the switch configured?

nickirwin

4 Posts

July 23rd, 2014 14:00

The switches are HP Procurve 2900's. Flow Control is on, Spanning Tree is off. Yes, the iDRAC is plugged into the same switch which losses connectivity due to the entre switch being flooded. Ports are Auto Configured and we have various VLANs. The symptons are very similar to a network loop but do not go outside the switch the server is connected, so does not flood the entire Subnet/other switches. We know its the Broadcom's that are causing the issue as if we disable only those ports on the switch it all comes back to life.

nickirwin

4 Posts

July 23rd, 2014 15:00

Couple of other notes...... * If we disable the VLANs on the switch ports the broadcoms are plugged into, the problem goes away * We also use Intel network cards in the same server that do no cause an issue * Other servers (R610's) with Broadcom's (not sure on model) do not have the issue It almost seems like an odd compatibility issue between a R720, a Broadcom 5720, a HP 2900 switch and VLANs. Our only other thought is a loop is being created within the Broadcom's itself upon a reboot where we would like to do some testing with Spanning Tree. We have left Flow Control on as this is what is recommended by VMWare.

Anonymous

5 Practitioner

 • 

274.2K Posts

July 24th, 2014 06:00

Are all 4 ports placed in the same VLAN? You might be right about the spanning tree. Does the switch show anything in the logging about spanning tree topology changes? Is portfast enabled on the ports that connect to the 5720 ports? If portfast is enabled, might try disabling it. Once the servers are up and running and everything seems to be working, check the status of the ports, see if any are in a blocking/discarding status.

nickirwin

4 Posts

July 28th, 2014 08:00

A combination of removing the vSphere management ports from multiple VLANs and turning on Rapid Spanning Tree seems to have resolved the issue.

 

bolson88

1 Message

August 19th, 2014 15:00

I have 10 R720 servers purchased in June of this year with the exact same issue. I've opened up a ticket with Dell Enterprise Support and they're going to try to re-create the issue.

dellserver12

1 Message

September 26th, 2014 10:00

There was a fix in the latest driver\firmware release on Sept 9, 2014 (7.10.x) that might address this issue.  Suggest you download and test this version.

View More

View All

No Events found!

Top