Proxmox VE40 incompatible with HP Proliant Servers...

[p1v12002]

after a month of trying I checked out the incompatibility of the new release with HP Proliant server (in my case DL385 G8 and DL360 G7). After service pack & firmware updates, an error occurs in the raid controller installation that causes a server failure. The HP Support support is limited to say that Debian is not among the supported OS (About to remind you that the previous Proxmox VE 3.4 version works properly...)
Many thanks in advance for any suggestion
Regards
Paolo

 

[manu]

Hi Paolo
Can you post a screenshot, of even better of a video of what went wrong, and what is the exact model of the raid controller ? ( like lspci output from a live CD=

Also to rule out kernel issues, did you try to install a standard Jessie 8.2 ?
I know HP uses Debian a lot internally so I would be surprised if we can't work this out ...

 

[p1v12002]

Hy Manu,
the problem is exactly as following description:

http://h20566.www2.hpe.com/hpsc/doc...363_OptionsProLiantServersStorage_critical__/

In details, I'm trying with a HP Proliant Server DL385 G8 Server witch Raid disk controller p420i; no differences even after the latest SPP2015.10.0 firmware upgrade: during the installation, the fans run at maximum speed and the server crashes with error reporting on the raid controller.

Many thanks in advance
bye
paolo

 

[Q-wulf]

stupid question:
Any chance the SmartArray Conroller is broken ?
I had a similar Issue with a Areca Raid-Controller and it turned out it was a broken controller.

 

[adamb]

We aren't having any install type issue with 4.0 on DL 380's. I haven't done a bunch with the Gen8 and Proxmox 4 but it did install if I recall. Gen9's have no issue what so ever.

 

[anigwei]

Hi!

We are having the same issue in a DL385 G8

As says p1v12002, the Proxmox installation hangs with fans at 100%. Later I tried to boot with kernel args "iommu=soft nmi_watchdog=0" and it boots correctly, but it get stuck at 0% when partitioning.

In a fresh Debian 8, everything is OK.

If I try to install Proxmox from Debian, and reboot with new kernel (*-pve), it hangs just after initial boot.
If I re-reboot the "proxmoxed Debian" with its original kernel, all is fine then.

BTW: I have blacklisted the module hpwdt as I had similar problems related to Watchdog and proliants... (Thread: 24015-VE-4-0-Kernel-Panic-on-HP-Proliant-servers in this forum)

Greetings.

 

[manu]

Hi
Unfortunately I don't have access to this hardware in the lab, so we need your help to pinpoint the issue
Since you already have Debian Jessie installed, can you install the debian package linux-image-amd64 from Debian backports ( it is in version 4.2+68~bpo8+2 as of today ) and see if it boots properly ?
You should be fine with disabling the hardware watchdog, in this case we will default to a software watchdog.

 

[anigwei]

Hi
Unfortunately I don't have access to this hardware in the lab, so we need your help to pinpoint the issue
Since you already have Debian Jessie installed, can you install the debian package linux-image-amd64 from Debian backports ( it is in version 4.2+68~bpo8+2 as of today ) and see if it boots properly ?
You should be fine with disabling the hardware watchdog, in this case we will default to a software watchdog.

Hi Manu!

I'm very sorry, but we needed the cluster running and we have ended downgrading all the nodes to Proxmox 3.4 and recreating the cluster, where everything runs smoothly . So that machine is not available now

If you want I can test with a Proliant 320 G6, which theoretically was failing too due to watchdog module.
The pity is that with DL385 G8 the error is immediate at boot/installation and with 320 G6 the error appears randomly during operation.

Andreu

 

[manu]

Hi Andreu
Yes we're insterested at this issue and also the watchdog issue.

> If you want I can test with a Proliant 320 G6, which theoretically was failing too due to watchdog module.

Do you mean you had to blacklist hpwdt.ko here ?

 

[anigwei]

Hi Manu,

Yes

* HP Proliant DL320 G6 install and run fine. But strange and random hangs with this ILO LOG:

An Unrecoverable System Error has occurred (Error code 0x0000002B, 0x00000000)​

* HP Proliant 320 G8 we had no hpwdt issues. Installation and runtime fine.
* HP Proliant DL350 G7 no hpwdt issues and installation & runtime fine.

* HP Proliant DL350 G8 we had *LOTS* of problem.
>* Installer gets hung with fans at 100% RPM. "Solution": Add iommu=soft at grub options
>* After installation, core dumped before Login screen: Add nmi_watchdog=0 at grub option
>* Boot with a LiveCD, blacklist hpwdt.so. Then only needs iommu=soft at grub options
>* After all that... weird HP Smartarray i420 issues...: https://forum.proxmox.com/threads/23781-Proxmox-VE-4-0-released!?p=122580#post122580 and http://forum.proxmox.com/threads/24793-Proxmox-VE40-incompatible-with-HP-Proliant-Servers).

* I lost any hope and returned back to Proxmox 3.4 (I had also weird issues related to the famous "401 and Ticket permissions". (My thread just here: https://forum.proxmox.com/threads/24889-Issues-in-cluster-on-Proxmox-v4).


Regards!

​

 

[NickTheGreat]

Hi Manu,

I have a proliant DL 385p Gen8 that I am running into the same problem with. It is not currently in production, so I can try to work with you to troubleshoot this further. I don't have anything installed on it yet, would you like me to try installing Debian and going down that road, or work directly with the proxmox installer?

I can get you a video of the install as well if that is helpful. For the record, so far my experience seems to be the same as the others (pve 3.4 installs fine, 4.0 and 4.1 do not, either of those installs hangs and fans begin to run at full speed, capslock/numlock don't work)

I have attached a picture of the last text I see on the screen when the system hangs.

-Nick

 

[NickTheGreat]

Hi Manu,

Quick update, my system may not be of much help any longer. I reset the BIOS settings back to system default, and then went to advanced settings and told the system to use the backup BIOS, it was version A28 from 02/06/2014. The system was able to boot, but in the proxmox install, it only showed the usb thumb drive I was using for installation, I was not able to select the logical drive on the p420i controller. Just to test, I went back in to the BIOS and went back to the latest BIOS version, still say A28, but with a date of 09/03/2014. Booted to proxmox again, did not lock/freeze again, but again did not show the logical drive on the raid controller, only thumb drive I am using to boot and install.

-Nick

 

[Roko26]

hi everyone,

I have three appliances ML385p G8 with smart array p420i, I would like install Proxmox in some them, but I'm into the same case, (The bios has corrupted and fans begin to run at full speed).

I tried with Debian 8 jessie 8.2.0 and and all is well, but with debian packages-linux-image-amd64 version 4.2+68~bpo8+2, the problem as like as Proxmox 4.0 or 4.1 (2f9650d4-21).

I tried with bios default (A28 11/02/2013) and with bios upgrade (A28 09/03/2014), so the Smart Array firmware upgrade of version 5.42 to 6.64, in both cases is equal.

In debian Jessie 8.2.0 when all is well, the controller used in kernel linux 3.16.7 is hpsa driver.

when I begin the installation Proxmox 4.0 or 4.1 crashes before turning on the fans to 100%, therefore I don't know what controller for the array 420i use proxmox.

I think that the problem is the driver for smay array controller p420i used in proxmox

is there some best p420i driver for proxmox 4 or 4.1 ?

someone could help me giving me some idea to solve this case ?

thank.

 

[ufuk]

Hi;exactly same error with my hp proliant dl180 g9 with p440

 

[Roko26]

Then, I can conclude that Proxmox is incompatible with the ProLiant DL385 G8 servers with array controller p420i ?????

 

[fireon]

RedHat run in same issues. Know these problems with DL Series and Linux. We use HP Proliant Servers a lot of years, since G5. DL Series are not really useable. Yes it is working when you are lucky and you have the right firmwareversion. So when you have an DL Server that is working, never, never upgrade the firmware!

We use alway ML350 (also in newest G9), with the first G9 Server we had an problem to install proxmox, there was an to old parted from debian. But since 3.4 and this time 4.0, 4.1 we never hat a problem with.

 

[spirit]

For the "pci uncorrectabble error", it seem to be a firmware bug:

http://community.hpe.com/t5/ProLian...h-uncorrectabl-PCI-express-error/td-p/5995669

Please upgrade or contact HP support.

 

[Roko26]

I've been to contact HP support, for the problem ML385p G8 with smart array p420i. they got me a new bios and firmware upgrade version for the smart array, I upgraded with new version, just now it's smart array firmaware version 6.68, but the problem is the same when I start the install of proxmox 4.0 or 4.1 then the fans work to 100% and the system crashes with a messages of corrupt bios.

 

[spirit]

I've been to contact HP support, for the problem ML385p G8 with smart array p420i. they got me a new bios and firmware upgrade version for the smart array, I upgraded with new version, just now it's smart array firmaware version 6.68, but the problem is the same when I start the install of proxmox 4.0 or 4.1 then the fans work to 100% and the system crashes with a messages of corrupt bios.

can you install debian jessie then proxmox kernel, and see if you have also the same bug ?

 

[Roko26]

yes, I installed Linux version 8.2 and all goes well, but when I installed Linux-image-amd64 from debian backport all is wrong, exactly debian packages linux-image-amd64 version 4.2+68~bpo8+2, the problem as like as Proxmox 4.0 or 4.1