Port Trunking for ProCurve 2510G for use in esxi 5.1

[Andrew79]

Hello! 

I need your help

I have a Proliant DL380G5 server with two NICs. This server is installed ESXi 5.1. Network monitors on display at one switch and the switch type in the parameters set ip balancing hash.

These two ports on the server I connect to the switch ProCurve 2510G-24. And here I have a problem starts.
Just trying to unite them into a static trunk, command
"trunk 11,13 trk1 trunk", then tried to merge the trunk LCAP, using the "trunk 11,13 trk1 lacp". In both cases, the connection from ESXi sevrerom constantly falls off.

Now as a result of the ports 11 and 13 just included LCAP and not included trunk. Communication with the server is not falling, but the aggregation seems to not work (i think).

Please tell me how to configure the trunk to work with esxi?

Versin frimware HP:  Y.11.41, ROM N.10.02

Result command "show lacp":

                           LACP

   PORT   LACP      TRUNK     PORT      LACP      LACP
   NUMB   ENABLED   GROUP     STATUS    PARTNER   STATUS
   ----   -------   -------   -------   -------   -------
   4      Active    4         Up        No        Success
   6      Active    6         Up        No        Success
   8      Active    8         Down      No        Success
   11     Active    Trk1      Up        Yes       Success
   12     Active    Trk1      Up        Yes       Success
   13     Active    13        Up        No        Success
   15     Active    15        Up        No        Success
   17     Active    17        Down      No        Success

[Chrisd131313 · Accepted solution]

Hi Andrew,

Have you run through this doc relating to LACP and ESXi 5.1? It has examples for Cisco and HP switch connectivity.

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1004048

HTH

[Andrew79]

Thank you. Your link helped.

Link has become stable. Revealed the ability to make some routing problems with more :)
It remains an open question should be included in the LACP trunk or not.

[Chrisd131313]

How are you testing the link aggregation? If you sending data over a single session it is likely you will only see one of the links being utilised. With multiple sessions you will then see both links utilised. This is because ofthe way the trunks work, the algorithm is based on Source Destination MAC addresses. On the HP switch you could try enabling L4 load balancing - if you switch has this option...

trunk-load-balance l4-based

This might give you what you need, but in my testing with other Hypervisors I have not seen it taking affect, maybe in a VMWare world this will be different.

HTH

[Andrew79]

Check on WEB Console Switch under workload. It is quite normal distribution is now it suits me :)

Balancing on the switch will try to change, over time, to see statistics there is a difference or not.

Thanks for help.

[Chrisd131313]

No problem, please mark the topic as solved if this has helped you out.

[Andrew79]

Wanted to close the topic, but ...

A few days working with the trunk to LACP. There were two ports in the fall is the time when downloading large faylv on ESXi host a download finishes with the errors, and the port goes into a state of DOWN

   PORT   LACP      TRUNK     PORT      LACP      LACP
   NUMB   ENABLED   GROUP     STATUS    PARTNER   STATUS
   ----   -------   -------   -------   -------   -------
   3      Active    Trk4      Up        Yes       Success
   4      Active    Trk4      Up        Yes       Success
   5      Active    Trk5      Up        Yes       Success
   6      Active    Trk5      Up        Yes       Success
   7      Active    Trk2      Blocked   No        Failure
   8      Active    Trk2      Blocked   No        Failure
   9      Active    Trk3      Up        No        Success
   10     Active    Trk3      Up        No        Success
   11     Active    Trk1      Up        Yes       Success
   12     Active    Trk1      Up        Yes       Success
   14     Active    Trk6      Up        No        Success
   16     Active    Trk6      Up        No        Success

 Now back to the state trunk "trunk". After watching a couple of days, and then closed the subject.

Apparently esxi 5.1 does not work properly s LACP, or in this case.

[Chrisd131313]

So even though VMWaer claim MACP works with ESXi 5.1 there are a few bugs which cause it to fail. Thanks for the update.

[Andrew79]

I think this topic can be closed.

Trunk mode works fine. Only after the change of the trunk had a problem with a managment interface. Full reset interface solved the problem.