VMware Cloud Community
lpanunzi
Contributor
Contributor

after install esx 4.1 ... in the console a red line

Hello,

after installation completed esx4.1 in the display i can see an additional line in RED:

Proliant bla bla bla

Console

http://blablablalb

0:00:00:50:879 cpu2:4711)intrvector:290:Out of interrupt vectors

can you help me?

T hanks

Reply
0 Kudos
44 Replies
gfishbein
Contributor
Contributor

Good luck, but the cards definitely have issues. Vmware confirmed this for us and said they will make a note of the issue on the hcl (card sending to many interrupts, 4 per port).

It is safer to go for a card that is 100% then wait for a fix.

Reply
0 Kudos
dpkenn
Contributor
Contributor

We ran into the same issue with ESX4.1 - vMotioning on a HP DL 360 G7 and using the NC375T NICs. What we found is an article that addresses this specific issue with an update to the nx_nic driver. ESX 4.1 ships with 4.0.550 driver version for nx_nic. The update is at 4.0.570. From the console...

cat /proc/vmware/version | grep nx_nic

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=102602...

(note that there is a link under solutions that points to the ESX4.1 driver download page)

When we first ran across the problem, we had issues vMotioning VMs (especially on VMs >1GB). We would see errors like...

"A general system error occurred: Migration failed while copying data. Already disconnected."

"Migration to host <<unknown>> failed with error Already disconnected (195887150)."

So we switched to the onboard NICs (NC382i) for vMotioning. Everything worked great. Switch back to the NC375T and things would fail again. Once we updated the driver and rebooted. Everything worked great. We could even put the hosts in maintenance mode and evacuate all VMs running without a single error.

-DK

Reply
0 Kudos
WellsITman
Contributor
Contributor

I have been running my own test for the last few weeks and even though you get the console errors everything seems to work fine on ESX 4.1 and the NC375T NICs (I am running driver 4.0.570). VMWare dont seem have made any comment on the HCL as they told gfishbein they were going to. Seems be to a good old case of Software vendor blames the Hardware vendor, and vice-versa.

Reply
0 Kudos
aty
Enthusiast
Enthusiast

Hi @ll,

I'm running into the same problems with our new HP DL580 G7. We also choosed two additional 375T quad-port cards to connect all the different vlans.

After reading this thread I checked the driver version and it is still the 4.0.550.1 (build 260247). Next thing to do is the update of that damned diver.

Thanks for the hints so far.

Greets

Reply
0 Kudos
WellsITman
Contributor
Contributor

Aty,

The latest version of the driver is 4.0.570.0 here http://downloads.vmware.com/d/details/esx4x_qla_nx_nic_dt/ZHcqYmRAdyViZCVodw==

Also HP now have a customer advisery about the issue with a work around and a note saying the issue will be fixed in the next driver. http://bizsupport1.austin.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSe...

We have been running Live in Production for 2 weeks now with 2 x DL580 G7 each with 2 x NC375T and have not had any issues.

Reply
0 Kudos
aty
Enthusiast
Enthusiast

Thanx WellsITman,

exactly that driver I have installed now. I can confirm that the error message on the console screen is still there, so whatever the driver 4.570 does, it is obviously not enough.

Since we do not use these nics for vmotion, I hope we get along untill the solution is available.

Reply
0 Kudos
gfishbein
Contributor
Contributor

I spent days on the phone with vmware and HP... the fix wont be released for a while.. contact your vendors and replace the nics for a supported version.

Reply
0 Kudos
WellsITman
Contributor
Contributor

Did you not read the HP customer advisery? It tells you how to resolve the error

http://bizsupport1.austin.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSe...

Reply
0 Kudos
wharlie
Contributor
Contributor

I have 2 x new DL580 G7 ESXi 4.1.0 320092.

HP NC375i Integrated quad port

2 x NC360T dual port

The vmotion portgroup is setup to use the NC360T (with NC375i as standby).

During testing when I vmotion it often fails with error:

Migration to host <<unknown>> failed with error Already disconnected (195887150)

And occosianally the vm (2K3std) has bluescreened immediately after vmotion with error:

IRQL_NOT_LESS_OR_EQUAL

I am running ESXi and don't see the console error of the OP, is there another way to check for the interupt vector error on ESXi.

Reply
0 Kudos
gfishbein
Contributor
Contributor

It's a "workaround" not a fix....

for us these are production servers, and a workaround is not a solution i would trust.

must safer to replace the nics with supported cards that don't require a "workaround"

Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

Hmm I have this issue too and both http://bizsupport1.austin.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSe... and http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=102602... seem to contradict each other.

One says to work around, the other says to use the newer driver.

Which is right for 4.1 please?

Reply
0 Kudos
WellsITman
Contributor
Contributor

Hutchingsp,

All I can tell you is using the HP work around with the 4.0.570 driver has been working fine on our new servers. They worked fine in testing for 3 weeks and have been in production for 6 weeks and have not had any issues.

Reply
0 Kudos
admin
Immortal
Immortal

Did you guys use the VMware ESXI download or the HP OEM Download direct from HP as I understand this driver is now included..

Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

ZippyDaMCT wrote:

Did you guys use the VMware ESXI download or the HP OEM Download direct from HP as I understand this driver is now included..

In my case (so far) the ESXi download - though we're not really beyond having installed it and configured MPIO so I could easily enough download/install the HP product if there is a benefit beyond downloading ESXi 4.1 from vmware and adding the driver download from vmware using esxupdate.

Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

ZippyDaMCT wrote:

Did you guys use the VMware ESXI download or the HP OEM Download direct from HP as I understand this driver is now included..

In my case (so far) the ESXi download - though we're not really beyond having installed it and configured MPIO so I could easily enough download/install the HP product if there is a benefit beyond downloading ESXi 4.1 from vmware and adding the driver download from vmware using esxupdate.

Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

ZippyDaMCT wrote:

Did you guys use the VMware ESXI download or the HP OEM Download direct from HP as I understand this driver is now included..

In my case (so far) the ESXi download - though we're not really beyond having installed it and configured MPIO so I could easily enough download/install the HP product if there is a benefit beyond downloading ESXi 4.1 from vmware and adding the driver download from vmware using esxupdate.

Reply
0 Kudos
hutchingsp
Enthusiast
Enthusiast

ZippyDaMCT wrote:

Did you guys use the VMware ESXI download or the HP OEM Download direct from HP as I understand this driver is now included..

In my case (so far) the ESXi download - though we're not really beyond having installed it and configured MPIO so I could easily enough download/install the HP product if there is a benefit beyond downloading ESXi 4.1 from vmware and adding the driver download from vmware using esxupdate.

Reply
0 Kudos
WellsITman
Contributor
Contributor

Hi All,

For those who have not already noticed, 4.1 U1 has just been released and it specifically fixes this error. From the release notes;

ESXi hosts might not boot or might cause some devices to become inaccessible when using NetXen 1G NX3031 devices or multiple 10G NX2031 devices.
When using a NetXen 1G NX3031 or multiple 10G NX2031 devices, you might see an error message similar to the following written to logs of the ESXi 4.1 host after you upgrade from the ESXi 4.0 host: Out of Interrupt vectors. On ESXi hosts where NetXen 1G and NX2031 10G devices do not support NetQueue, the ESXi host might run out of MSI-X interrupt vectors. This issue can render ESXi hosts unbootable or other devices (such as storage devices) inaccessible. The issue is resolved in this release

Reply
0 Kudos
r_3__
Contributor
Contributor

Just wanted to update everyone on this chain of issues, according to VMware's HCL the NC375T (for 4.1 & U1) is not listed.  So rather than fixing the issue, they removed it from the supported hardware list.  I have several of my clients with HP systems running the NC375T cards and most have not had an issue but the question is, if they do have issues, will they get support?  Thanks VMware and HP!!!! 😞

Reply
0 Kudos
WellsITman
Contributor
Contributor

What? I can still see it in the HCL for ESX 4.1 U1. It says it has inbox support for it with driver 4.0.550.1. But it does still have this problem, its not fixed like they claim in the U1 fix list - this issue is specifically mentioned in the releasenotes as being fixed, but it still isnt.

http://www.vmware.com/resources/compatibility/search.php?action=search&deviceCategory=io&productId=1...

Reply
0 Kudos