Hello,
I have a VMware ESXi 5.0.0, 623860 host running two Windows VM's pretty well. The host hardware is an HP Proliant DL 380 G5 server.
This host is GBit Ethernet connected to an HP Procurve 2810-48G (J9022A) switch via two Cat 5e UTP cables.
I want to configure a ethernet trunk using the two ports of switch for fault tolerance and link failover disaster.
I telnet the switch and configure the ports as Trunk3 with TRUNK option (not LACP) as shown in this pic:
At the ESXi side I go to the vSphere Clent at the Configuration --> Networking --> vSwitch0 --> Propreties --> Edit --> NIC Teaming and set the Load Balancing option to "Route Based on IP hash" as shown in the next pic:
At this point with or without reboot the Procurve or the ESXi host I can ping the host and the two virtual machines in it, but after about 10/15 minuts I loose the connectivity to the host (ping stops). Then if I disconnet one of the UTP cables the connectons resumes.
I'm not sure if those are the right configurations for my purpose, so any help would be very appreciated.
Thaks a lot.
OLIVEIRAS
Hi Kamen (and oliveiras - in case you haven't found a solution to this yet).
I had the same problem recently with ESXi 5.1 and an HP switch and I fixed it by following this link:
The setup seems good with a quick look.
Could you enable CDP on the vSwitches to make sure the right vmnics maps to your physical ports:
http://rickardnobel.se/archives/1110
and then post the result from "show lldp i r" on your 2810.
Hi Rickard,
As you said:
"The setup seems good with a quick look.
Could you enable CDP on the vSwitches to make sure the right vmnics maps to your physical ports:
http://rickardnobel.se/archives/1110
and then post the result from "show lldp i r" on your 2810."
This is the result of the show lldp command on my procurve switch:
The problem remais, but now I lost the ping to the host (connectivity) immediately. Then if I disconet one of the cables the connection returns.
What am I missing here?
Thanks a lot.
Everything looks good from what I see, and through the CDP we can be sure that the cables are correctely connected. The trunk configuration seems fine too.
Could you run these commands and paste the results?
as mentioned by rickard almost every thing is looking normal
just for checking... any jumbo frames using in the switches, any firmware updates for the pNICS or HW issues, may be VM related.?
Here are the results of the commands on my Procurve 2810-48G. I must say that now, after the previous post, I need to have one UTP cable disconected to have connectivity.
show interface 3-4
show interface config (the full sugested command resulted on error "invalid input: 3-4")
Thanks again for the help.
OLIVEIRAS
Everything looks very good on your switch, nothing seems to be bad or incorrectly configured.
On the ESXi host, what portgroups are on the vSwitch?
If you go to the settings on these portgroups, are there anything configured directly on those? Like some override on the NIC Teaming policies? Or any active / standby vmnics?
I notice that CDP (Cisco Discovery Protocol) is not available.
The hardware host is an HP Proliant DL380 G5 and the TEAM nic's in question are "Broadcom Corporation NC373i Integrated Multifunction Gigabit Server Adapter"
Any NIC driver update that enables CDP? Is there the CDP option thar is needed to this to work?
Thanks again.
OLIVEIRAS
I found those links by googling. I'm not sure if the LLDP and CDP options are exclusively for the paid versions of ESXi?
http://wahlnetwork.com/2012/07/17/utilizing-cdp-and-lldp-with-vsphere-networking/
http://www.ivobeerens.nl/2012/05/16/configure-the-link-layer-discovery-protocol-lldp-in-vsphere-5/
OLIVEIRAS
I want to configure a ethernet trunk using the two ports of switch for fault tolerance and link failover disaster.
This may not answer your question about how to setup the trunk, but I wonder why you want to do this at all? With the default settings "Route based on originating port id" and the physical ports configured as as untagged (or tagged, depending on your needs) ports you already have all you want with the two uplinks.
André
oliveiras wrote:
I notice that CDP (Cisco Discovery Protocol) is not available.
You have already CDP enabled, see my link earlier. What you see above is incoming CDP, which is not available for you.
Ok CDP is enable but when I enable the two ports trunk in my Procurve switch after about 6 minuts I loss the ip connectivity to my ESXi 5 host but not to the VM's on it.
Note: Using the defaults configs for VMware Host Networking.
OLIVEIRAS
Here it seems like port 5 has also been added as a third member of the Trunk? Where does this port attach?
Hi Rickard,
It was my mistake here. But not working with the correct ports trunk configured.
OLIVEIRAS
The management portgroup, any settings on that? As some active/standby or other override?
Hello OLIVEIRAS,
what was the solution in your case?
I have pretty much the same issue.
It would be great if you share it with the community.
Thanks,
Kamen
Hi Kamen (and oliveiras - in case you haven't found a solution to this yet).
I had the same problem recently with ESXi 5.1 and an HP switch and I fixed it by following this link:
Hi,
It seems to solve the problem here.
Thanks a lot!