Lost WOL support on BCM5708 after ESX 3.5 update

[RS_1]

Hi,

I was about to test DPM feature on a cluster (7x IBM HS21 8853 C2G) when i notice a strange thing.

One of the 7 blade was not up to date (ESX 3.5.0 110268) but the others where (3.5.0 120512). On the 6 up-to-date blade, the wake on lan support is "no" but on the out-of-date one the WOL support was "yes"

I updated and guess what...

Any one could explain that ?

Thx

[RS_1]

The same adapter on a HP DL580 G5 :

BTW, as IBM says here :

"The WoL function operates correctly. A hardware issue in the affected chip prevents the ethtools code from correctly querying the WoL function, and, therefore, it reports that WoL is disabled and not supported"

[RS_1]

same issue with Update 3 :

[calvinso]

I have the same problem after upgrading from U2 to U3, all my HP servers lost WOL Support. Which it works in U2.

[RS_1]

So it's not an IBM issue.

VMware told me the same thing today but i was really not sure about that.

Did you ask support for that ?

[calvinso]

I have place online support, he just reply me a silly KB , which telling me that my nics is tested by VMware and should support WOL.

I don't know what is the purpose for him send me such KB, Of cz I know my NICs have WOL, I am telling them U2 is working but U3 fails. I am really getting angry with VMware supports.

By the way, I have place 3 technical supports on Vmware, none of them are solved and repled with sense. Each reply need to wait at least 3 days - 7 days while those replies are non-sense replies. Nothing helping the case. I am wondering why their support level is so suck.

[RS_1]

Really helpfull...

i'll ask the support too and tell you what they told me.

[calvinso]

I have new findings

U3 report BCM5708 as fiber lan card !

# ethtool vmnic0

Settings for vmnic0:

Supported ports:

Supported link modes: 1000baseT/Full

Supports auto-negotiation: Yes

Advertised link modes: 1000baseT/Full

Advertised auto-negotiation: Yes

Speed: 1000Mb/s

Duplex: Full

Port: FIBRE

PHYAD: 2

Transceiver: internal

Auto-negotiation: on

Supports Wake-on: d

Wake-on: d

Cannot get message level: Function not implemented

Link detected: yes

While U2 report it as MII Twist Pair

# ethtool vmnic0

Settings for vmnic0:

Supported ports:

Supported link modes: 10baseT/Half 10baseT/Full

100baseT/Half 100baseT/Full

1000baseT/Half 1000baseT/Full

Supports auto-negotiation: Yes

Advertised link modes: 10baseT/Half 10baseT/Full

100baseT/Half 100baseT/Full

1000baseT/Half 1000baseT/Full

Advertised auto-negotiation: Yes

Speed: 100Mb/s

Duplex: Full

Port: Twisted Pair

PHYAD: 1

Transceiver: internal

Auto-negotiation: on

Supports Wake-on: g

Wake-on: g

Current message level: 0x000000ff (255)

Link detected: yes

# ethtool

While I try to type "ethtool -s vmnic0 port mii"

# ethtool -s vmnic0 port mii

Cannot set new settings: Function not implemented

not setting port

#

[RS_1]

what's your HP hardware type/model ?

because if it's blame, the FIBER parameter is good (it's see as FC between blade and FC switch within the chassis)

[trico]

Hi,

I do have the same problem on my HP BL460c G1.

After applying U2 post fix "Update Software Broadcom Driver" on ESX3.5, I lost WOL.

Same on ESXiU2 with patches before U3.

Same with fresh instal of ESXiU3.

Does anyone already get an answer from VMWARE

Thanks

[NickTr]

Hi,

I noticed the same thing with my HP BL 460c G1 servers. I actually noticed the problem when installed the non-critical host updated prior to U3 being released. I believe the issue was to do with a Broadcom software update released on the 18th September (). As I had not realised I had lost WOL I installed it across 16 of my ESX servers. I then tried using DPM and that is when I realised it was no longer working. As there is no way to uninstall a patch I contacted HP support (VMware supported through HP). Eventually they supplied me with a new development NIC driver in a rpm. This did enable the WOL feature on the broadcom NIC and appeared to resolve the problem!

I did find that the WOL feature worked shortly after a host had been put into standby (by DPM or manually) but stopped working when the host was in standby for a long period of time (e.g. a day). I still haven't got to the bottom of this.

When installing U3 the same thing happened again so I am guessing the broadcom update has been rolled into U3.

I'd rather not go through all my hosts again and manually install the new driver rpm so I'm hoping VMware will release a patch.

Cheers,

Nick

[calvinso]

I got this reply finally from VMware support

"Explanation as per Broadcom Engineer:

---

The hardware that you have with the 5708 rev 12, corresponds to the B1 revision of the 5708. The rev 12, corresponds to the B2 revision of the 5708. In the bnx2 code, we have a section of code in the function which will disable WoL for 5708 B1 revisions. This is done because the driver can't reliably control the WoL settings. This should explain why WoL is disabled. They also checked bnx2 v1.5.11 in Linux code and it has the same code above which disables WOL on rev11. However, they removed the WoL feature from this version of the driver. They also checked bnx2 v1.5.11 in Linux code and it has the same code above which disables WOL on rev12.

In conclusion this a driver bug and is yet to be fixed by Broadcom.

By the way, I think VMware is doing bad here, they should have notice that before issuing update and include it in release notes!

"

[calvinso]

can you upload the rpm to somewhere and let us have a testing to?

[RS_1]

would you take the risk to apply this rpm in production ? :smileygrin:

thanks for sharing information BTW.

[calvinso]

I really want to try and I will install on 1 of my 4 VM hosts to test. (Actually I even planning to downgrade it back to U2 =.=")

[RS_1]

you really use DPM that much ?!

Le 5 déc. 08 à 02:07, "calvinso" <communities-emailer@vmware.com> a

écrit :

raphael,

A new message was posted in the thread "Lost WOL support on BCM5708

after ESX 3.5 update":

http://communities.vmware.com/message/1116260#1116260

Author : calvinso

Profile : http://communities.vmware.com/people/calvinso

Message:

[calvinso]

previously it will shutdown 6 server out of 8 at night, suppose each blade server using around 400VA (all 10 fan speed will decrease from 60%-&gt;31% too), how much running cost we could save? (power, air condition, etc.)

[RS_1]

A lot off course...

[calvinso]

This problem have been fixed in 3.5 Update 4.

[RS_1]

Yes, i tested it today