Problem with kernel 2.6.35 and Qlogic FC driver on IBM Bladecenter

gkaraolides

New Member
May 25, 2011
8
0
1
Greetings,

I am using ProxMox VE 1.8 on servers in an IBM Bladecenter. Here is the output of pveversion -v:

--- Begin pveversion -v ---

gendebien:~# pveversion -v
pve-manager: 1.8-18 (pve-manager/1.8/6070)
running kernel: 2.6.35-1-pve
proxmox-ve-2.6.35: 1.8-11
pve-kernel-2.6.32-4-pve: 2.6.32-33
pve-kernel-2.6.35-1-pve: 2.6.35-11
qemu-server: 1.1-30
pve-firmware: 1.0-11
libpve-storage-perl: 1.0-17
vncterm: 0.9-2
vzctl: 3.0.28-1pve1
vzdump: 1.2-14
vzprocps: 2.0.11-2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.14.1-1
ksm-control-daemon: 1.0-6

--- End pveversion -v ---

Each blade has a Qlogic FC card connecting to the bladecenter's Qlogic FC switches and the fabric is connected to an IBM Storwize v7000 FC-capable storage server.

With kernel version 2.6.32 everything works fine.

I tried kernel version 2.6.35 to get KSM functionality, but the FC would not come up. Here is the relevant output of dmsg:

--- Begin dmsg ---
qla2xxx 0000:24:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32
qla2xxx 0000:24:00.0: Found an ISP2532, irq 32, iobase 0xffffc9000067a000
alloc irq_desc for 72 on node -1
alloc kstat_irqs on node -1
qla2xxx 0000:24:00.0: irq 72 for MSI/MSI-X
alloc irq_desc for 73 on node -1
alloc kstat_irqs on node -1
qla2xxx 0000:24:00.0: irq 73 for MSI/MSI-X
qla2xxx 0000:24:00.0: Configuring PCI space...
qla2xxx 0000:24:00.0: setting latency timer to 64
qla2xxx 0000:24:00.0: Configure NVRAM parameters...
qla2xxx 0000:24:00.0: Verifying loaded RISC code...
qla2xxx 0000:24:00.0: FW: Loading via request-firmware...
qla2xxx 0000:24:00.0: Allocated (64 KB) for FCE...
qla2xxx 0000:24:00.0: Allocated (64 KB) for EFT...
qla2xxx 0000:24:00.0: Allocated (1350 KB) for firmware dump...
qla2xxx 0000:24:00.0: Unable to read FCP priority data.
qla2xxx 0000:24:00.0:
QLogic Fibre Channel HBA Driver: 8.03.02-k2
QLogic QMI2572 - QLogic 4Gb Fibre Channel Expansion Card (CIOv) for IBM BladeCenter
ISP2532: PCIe (5.0GT/s x4) @ 0000:24:00.0 hdma+, host#=0, fw=4.04.04 (80)
alloc irq_desc for 42 on node -1
alloc kstat_irqs on node -1
qla2xxx 0000:24:00.1: PCI INT B -> GSI 42 (level, low) -> IRQ 42
qla2xxx 0000:24:00.1: Found an ISP2532, irq 42, iobase 0xffffc900110b6000
alloc irq_desc for 74 on node -1
alloc kstat_irqs on node -1
qla2xxx 0000:24:00.1: irq 74 for MSI/MSI-X
alloc irq_desc for 75 on node -1
alloc kstat_irqs on node -1
qla2xxx 0000:24:00.1: irq 75 for MSI/MSI-X
qla2xxx 0000:24:00.1: Configuring PCI space...
qla2xxx 0000:24:00.1: setting latency timer to 64
qla2xxx 0000:24:00.1: Configure NVRAM parameters...
qla2xxx 0000:24:00.1: Verifying loaded RISC code...
qla2xxx 0000:24:00.1: FW: Loading via request-firmware...
qla2xxx 0000:24:00.1: Allocated (64 KB) for FCE...
qla2xxx 0000:24:00.1: Allocated (64 KB) for EFT...
qla2xxx 0000:24:00.1: Allocated (1350 KB) for firmware dump...
qla2xxx 0000:24:00.1: Unable to read FCP priority data.
scsi1 : qla2xxx
qla2xxx 0000:24:00.1:
QLogic Fibre Channel HBA Driver: 8.03.02-k2
QLogic QMI2572 - QLogic 4Gb Fibre Channel Expansion Card (CIOv) for IBM BladeCenter
ISP2532: PCIe (5.0GT/s x4) @ 0000:24:00.1 hdma+, host#=1, fw=4.04.04 (80)
hub 2-3:1.0: USB hub found
hub 2-3:1.0: 2 ports detected
usb 3-2: new full speed USB device using uhci_hcd and address 2
qla2xxx 0000:24:00.1: LOOP UP detected (4 Gbps).
qla2xxx 0000:24:00.0: LOOP UP detected (4 Gbps).
qla2xxx 0000:24:00.0: MRK-IOCB: Invalid completion handle (0) -- timed-out.
qla2xxx 0000:24:00.1: MRK-IOCB: Invalid completion handle (0) -- timed-out.

--- End dmsg ---

This problem appears to be the same as this, discussed on Qlogic's own support forum:

http://solutions.qlogic.com/KanisaS...d=show&forward=threadedKC&kcId=Post-265102001

Apparently a problem with the Qlogic FC driver arose with kernel version 2.6.35, and by 2.6.36.2 it was solved. It's possible that it was also solved in later fixes to 2.6.35 but I don't know.

It would be good to have KSM, but since all VM's and data here are on the FC SAN, kernel 2.6.35 is a non-starter.

If anyone has determined a solution or a workaround please post it. Otherwise please have this in mind.

With best regards,

George Karaolides
 
Greetings,

Have tried pve-kernel-2.6.35-2-pve_2.6.35-12_amd64.deb and Fibre Channel works.

Unfortunately there are now problems with the networking of the virtual hosts.

We have other blades in the same Bladecenter running the same ProxMox software, where the Fibre Channel did not show problems with pve-kernel-2.6.35-1 (possibly because of different model Qlogic fibre channel cards).

I tried both pve-kernel-2.6.35-1 and pve-kernel-2.6.35-2 on these machines, and the problem with the virtual host networking appeared with both kernel versions. So this is not a problem introduced with pve-kernel-2.6.35-2. It also existed on 2.6.35-1 but I only noticed it once the Fibre Channel problem was solved.

With best regards,

George Karaolides
 
Sorry, I replied to this a few days ago but somehow my reply wasn't posted and I didn't notice - something to do with the forum session timing out and me having to log in again. And in the meantime we have been facing issues with power supply following the explosion yesterday that destroyed the biggest power station here on Cyprus - you may have heard it or read about it in the news.

The problems with the networking of the virtual hosts when using kernel 2.6.35-1 and 2-6-35-2 are that, on rebooting the virtual host, initially it is possible to get some erratic and unreliable networking, and then network access is lost.

For example, after rebooting the virtual host, it is possible to make an ssh connection to it, but it is not possible to query the DNS server from the virtual host (DNS server in the same LAN). On trying to ping the DNS server from the virtual host, the SSH session becomes unresponsive; the ping errors continue to appear on screen but the session will not respond to keyboard input. Eventually all network access is lost.

Best regards,

George
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!