HP DL360G4, Broadcom drivers, and EL4 Update 4 drop off network

[Tom Garske]

I upgraded the broadcom drives, however, the problem still remains. How did you update the IMPI firmware when they require you download and execute the softpaq on a windows platform?
RHAS 4.0

[Sergio J. Larrondo]

There's a message above from Chad Slater with a link I followed and I used the DOS update. So far no hiccups with the firmware/new driver combination, but I'm not 100% sure just yet.

[Tom Garske]

Nevermind, I cant read

[Chad Slater]

UPDATE:
I've been running the fix I recommended for a month without seeing the problem reoccur. I hope this works for those of you who can use it.

If you have a different Broadcom chip, perhaps there is an IPMI fix for that chip as well?

If have two NICs, and you are not using both NICs, you can try to connect to the second port. If it is an IPMI problem, that may be an acceptable workaround until permenant fix is available for your chip. It is also faster than updating the firmware.

[Berger Michel]

I have a similar problem on a ML 570 G2 running RH AS 3.0 2.4.21-37.0.1.ELsmp kernel with a Broadcom Corporation NetXtreme BCM5703 Gigabit Ethernet (rev 10) : each time we run a backup with HP Data Protector, after 20 mins the card hangs and we have to reboot the server (network restart does not help). Yesterday, we changed the card but it has happened again ! I have updated the firmware of the card like suggested and I installed the latest driver from HP :
driver: bcm5700
version: 8.3.14a
firmware-version: 5703-v2.35
bus-info: 05:02.0

Another strange thing is that we have a second identical server and we never had this problem with this one. For the time being, I don't know what I can try next.
Michel.

[Phil Crooker]

This is in reply to Michael Berger's comments. We too are in the same situation as we have the 5703 chip (NC7781 Gigabit Server Adapter).

The difficulty here is the servers with the 5703 chip is out of production so I don't know how much effort HP would be willing to put in. The firmware is dated July '03.

We've resorted to putting in a 3rd party card (Intel GB). Not pretty and not very practical if you have alot of these machines....

I just sent a form to the HP support group - I'll report on any response.

[Phil Crooker]

The HP techs recommend upgrading to the latest firmware - it turns out there is more up to date firmware for the DL380G3 machines than I had been able to find before.

The process is to first upgrade the BIOS to version (P29) 2004.09.15:

http://h18023.www1.hp.com/support/files/server/us/download/22034.html

(I don't know the DL360 models, here is the version for the DL360G2 - dos binary):

http://h18023.www1.hp.com/support/files/server/us/download/20885.html

Once the BIOS is done, then the NIC firmware to ver 1.2.3-1:

http://h18023.www1.hp.com/support/files/server/us/download/24045.html

Then the driver (broadcom 5700 ver 8.3.14a - you could of course just get this from broadcom...):

http://h18023.www1.hp.com/support/files/server/us/download/23968.html

We will be applying this to our relevant servers but we don't get enough traffic to turn the error regularly. If anyone does this who does get the error alot, can you please report back if it works - I want to post the results to the open RedHat and Debian bugs.

[Robert Walker_8]

Sergio et al,

On the 10/1/07 we upgraded the Firmware on our DL380 which was having problems and so far touch wood - 14 days since there have been no problems:

Extract from our inhouse system logging process:

Upgraded nic firmware (q57udiag -sil -firm 7782327b.bin & q57udiag -cname 7782 -uipmi pt5704c2.36) and reboot - if this should fail then turning off IPMI is another suggested solution (q57diag -c 0 -ipmi 0) using the nc7782.dsk image.

Regards,

Robert.

[Berger Michel]

In my case (ML 570 G2 running RH AS 3.0 2.4.21-37.0.1.ELsmp kernel with a Broadcom Corporation NetXtreme BCM5703 Gigabit Ethernet rev 10), I finally succeeded to convince HP to replace the mother board and I have no more problems with the Data Protector backup. So, the problem is solved but I have no real explanation.