kernel panic during 3.5 install

[nem8]

We are in the process of transitioning to 3.5 from 3.01. During install just seconds after hitting enter at the boot prompt I get a kernel panic. I have discovered what is causing the panic and need help finding a way around it. We are doing the install remotely using the ILO on a HP DL585 G2 and disableing the qlogic HBA cards so we dont accidently delete some stuff off the san.. My issue is that if we disable the qlogic HBA and try to do an install we get a kernel panic.. Re-enable the cards in bios and install starts just fine.. We need to remotely be able to disconnect the san from this server without having to involve our storage team. I dont understand why disableing the qlogic HBA's in the bios causes a kernel panic. Is there maybe some boot options/switches I can use to bypass this problem? I have several more servers to install and traveling to each location is out of the question and I really cant count on the staff at each location to pull san cables.

[lamw]

We actually saw same issue but it was not due to the FC HBA's but due to the intial release of ESX 3.5 and doing an inline upgrade from 3.0.2 u1 + some patches, after they released some additional patches and by patching right after the upgrade without rebooting it solved our kernel panic. Also this has also been solved by doing an inline upgrade to ESX 3.5u1, there's a KB articel about this kernel panic when you do upgrades for some situations and we were also using 585 G2. I would double check to make sure you're using the latest ESX 3.5u1 zip package to do the inline upgrade. There is an article out there about re-creating initrd to not load the QLogic driver modules while building the host, but I think that's only useful if you're rebuilding and you would have to re-create the ramdisk for this customization. If you zone/mask your switches, you could have a SAN/Switch Admin remove that masking/zoning to ESX Server tempoarily but again this will rely on external team as you would have from the storage admins either from removing the lun mapping OR pulling the FC cables.

[Texiwill]

Hello,

I would verify that the BIOS on everything is first up to date. I had some strange issues on my HP equipment until I made sure of that.

Unless you are boot from SAN the Qlogic BIOS (within the qlogic setup screens) is disabled by default or should be. If you are disabling the cards within the HP BIOS that is another matter. If its the later, the cards do not really disappear.... And could cause some confusion.

There is a kickstart script hanging about that will make it so that SAN devices do not show up but using it may not 100% work all the time.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[nem8]

We are doing a fresh install on this server and we just updated all bios/firmware on this machine. There has got to be some sort of workaround for this like a switch to ignore or something like that when at the boot prompt.. I just cant find anything. We are disableing the cards in the HP bios so it shouldnt care but it does. Any other suggestions would be great. And to clear things up we are using 3.5 update 1.

[Texiwill · Accepted solution]

Hello,

We are doing a fresh install on this server and we just updated all bios/firmware on this machine.

Ah good.

>There has got to be some sort of workaround for this like a switch to ignore or something like that when at the boot prompt.. I just cant find anything.

There is not one.

>We are disableing the cards in the HP bios so it shouldnt care but it does.

That is because a hardware scan shows the devices whether they are disabled or not. Actually you do not want to do this as much as you want to manipulate the install as it is happening. The best way is to start the install in Text mode, once in text mode let it go until get to the Welcome screen. Before you press OK use 'ALT-F2' to move to a different screen with a prompt. Run the command 'lsmod' then use 'rmmod drivername' to remove the driver. This in effect should take the qlogic card out of any further consideration for the installation. 'ALT-F1' will bring you back to the main installer screen.

I used the following to do this for my Emulex

rmmod lpfc_740

If you are using the graphical interface then use 'CTRL-ALT-F2' to get to the shell prompt and then use 'ALT-F7' to get back to the graphics.... Have not tried the last it could be different, you may have to run through ALT-F1 through ALT-F11 to find it however.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[nem8]

Texiwill,

We dont even get the the GUI portion of the install.. 10 seconds after hitting enter from the boot: prompt we get the kernel panic

[Texiwill]

Hello,

Do not disable the FC-HBA within the HP BIOS. That should solve the problem.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[nem8]

Correct.. Leaving the cards enabled does solve the problem, but according to vmware documentation you should not have the SAN showing during install so there is no way to erase/replace the data on the SAN. The only way to do that currently is to pull the HBA cables but I dont have confidence in the onsite people to pull the correct HBA cables and be able to plug them back in correctly.. This is why I am seeking a way to avoid any hands on solution.

[Texiwill]

Hello,

If you enable the FC-HBA in the BIOS and follow the instructions I listed, your FC-HBA will no longer be seen by the installer. This is the steps the kickstart script uses that I mentioned before..... Remember no driver loaded, no devices available.

You should get to the disk screen and only see local devices.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[nem8]

Texiwill,

Thanks for the advice it worked great I was able to get 3.5 installed remotely..

[Texiwill]

Hello,

Wonderful news. By way of thanks, you can award points within the forum for correct and helpful answers. This way members get a reward as well as thanks.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization