[gPXE] iSCSI boot works sometimes
Chad Voelker
chad.voelker at gmail.com
Tue Jan 25 21:21:30 EST 2011
To update the list on what I've done. I tried Andrew's suggestions
below, but entries in the blacklist file don't seem to make a
difference. I believe that this is either because a) the initramfs
process does not look at the /etc/modules directory or b) The scripts
mentioned in the link here:
http://etherboot.org/wiki/sanboot/ubuntu_iscsi2 do the loading of the
drivers explicitly. Though, if "b" was the explanation, I would expect
them to be loaded in the same order each time.
What I ended up doing was adjust this portion of the
/etc/initramfs-tools/scripts/local-top/iscsi file (hopefully, the
formatting comes across ok):
until ifconfig ${netdev}|grep -q "inet addr"; do
for netdev in `ifconfig -a|grep Ethernet|cut -d' ' -f1`; do
# Do a round-robin search for dhcp servers
ip link set ${netdev} up
# try to bring up the interface
ipconfig -t 2 -c dhcp -d ${netdev}
# Get an IP
if ifconfig -a|grep -q "inet addr"; then break 2; fi
# If we have got an address, stop searching.
# If we are here, then we have a problem, remove and re-add the NIC
rmmod forcedeth
# <--- Remove the NIC driver
sleep 1
# <--- Give it some time
modprobe forcedeth # <--- Re-add the NIC driver
sleep 2 # <--- Give it some time
done # end of netdev probing
CNT=$((${CNT} - 1))
if [ ${CNT} = 0 ]; then
echo "No dhcp servers found!"
exit 0
fi
echo "Tries left: ${CNT}"
done # end of until
I added the four lines below "If we are here, then we have a problem..."
Not the cleanest solution, but it has been reliable.
Thanks again Andrew for your help. If nothing else, you helped confirm
that I was looking in the right direction.
-Chad
On 01/19/2011 08:23 AM, Andrew Bobulsky wrote:
> Hello Chad,
>
> As a disclaimer, I've got next to no idea how device drivers on Linux
> work, so I've only got some guesses for you ;)
> I suspect you might be right about driver load order having something
> to do with it. Specifically, in your non-working example's dmesg
> output:
>
>
>> [ 1.249252] forcedeth 0000:00:0a.0: PCI INT A -> Link[LMAC] -> GSI 20 (level, low) -> IRQ 20
>> [ 1.249350] forcedeth 0000:00:0a.0: setting latency timer to 64
>> [ 1.302279] Loading iSCSI transport class v2.0-870.
>> [ 1.315354] forcedeth 0000:00:0a.0: ifname eth0, PHY OUI 0x732 @ 1, addr 90:fb:a6:2b:ee:ff
>> [ 1.315450] forcedeth 0000:00:0a.0: highdma csum pwrctl gbit lnktim msi desc-v3
>>
> I'm unsure as to whether the iSCSI driver used here cares whether or
> not a NIC is available before it goes hunting for the iBFT, but it
> could be possible. It looks like Linux kernels are designed to not
> care a whole lot about device driver load order... which seems odd to
> me, but I'm sure there's a reason for it.
>
> While the blog post I found is a little dated, the technique (though a
> little convoluted!) may likely still apply. Take a look here:
> http://rackerhacker.com/2009/01/26/linux-adjust-storage-kernel-module-load-order/
>
> Doubling down on that disclaimer again, the following suggestion is a
> complete guess:
>
> First, open your /etc/modprobe.d/blacklist file and add iscsi on to
> the end of it.
> Then, open your /etc/modules file and specify the forcedeth driver
> followed by the iscsi driver.
>
> Might work :P
>
> Best regards,
> Andrew Bobulsky
>
>
> On Wed, Jan 19, 2011 at 8:35 AM, Chad Voelker<chad.voelker at gmail.com> wrote:
>
>> Hello,
>>
>> I apologize for being slightly off-topic...
>>
>> I have a Ubuntu 10.04 Myth frontend that I'm trying to get booting with iSCSI. gPXE is working, and attaching the iSCSI drive and initiating the boot sequence on the iSCSI drive. Following the guide here: http://etherboot.org/wiki/sanboot/ubuntu_iscsi2, it sometimes fails on the file /etc/initramfs-tools/scripts/local-top/iscsi where it tries to call ipconfig to get DHCP info (even though the server is there and fine, otherwise we wouldn't have gotten this far).
>>
>> This works somewhere around 1 out of 3 boots. What I've noticed in dmesg is that when working, the iSCSI driver is loaded near the end of the process. I've listed (trimmed) versions of the dmesg output below, if you need the complete output let me know.
>>
>> Is there somewhere that I can dictate the order that things are loaded? Any other ideas?
>>
>> Thanks.
>> -Chad
>>
>> dmesg (entries that are repeated / unimportant are shortened with a "...")
>>
>> Working...
>>
>> [ 0.947158] udev: starting version 151
>> [ 1.115166] ahci ...
>> [ 1.132935] scsi0 : ahci...
>> [ 1.252750] ata1: SATA max UDMA/133 ...
>> [ 1.253462] forcedeth: Reverse Engineered nForce ethernet driver. Version 0.64.
>> [ 1.254782] ACPI: PCI Interrupt Link [LMAC] enabled at IRQ 23
>> [ 1.254869] forcedeth 0000:00:0a.0: PCI INT A -> Link[LMAC] -> GSI 23 (level, low) -> IRQ 23
>> [ 1.254965] forcedeth 0000:00:0a.0: setting latency timer to 64
>> [ 1.318176] forcedeth 0000:00:0a.0: ifname eth0, PHY OUI 0x732 @ 1, addr 90:fb:a6:2b:ee:ff
>> [ 1.318264] forcedeth 0000:00:0a.0: highdma csum pwrctl gbit lnktim msi desc-v3
>> [ 1.348043] usb 2-3: ...
>> [ 1.572051] ata1: SATA link down (SStatus 0 SControl 300)
>> [ 1.572193] ata2: SATA link down (SStatus 0 SControl 300)
>> [ 1.600025] ata6: SATA link down (SStatus 0 SControl 300)
>> [ 1.600139] ata3: SATA link down (SStatus 0 SControl 300)
>> [ 1.600146] ata4: SATA link down (SStatus 0 SControl 300)
>> [ 1.600202] ata5: SATA link down (SStatus 0 SControl 300)
>> [ 1.619530] Loading iSCSI transport class v2.0-870.
>> [ 1.667310] iscsi: registered transport (tcp)
>> [ 1.678221] iBFT detected at 0x93350.
>>
>> Not Working...
>>
>> [ 0.947224] udev: starting version 151
>> [ 1.247791] forcedeth: Reverse Engineered nForce ethernet driver. Version 0.64.
>> [ 1.249164] ACPI: PCI Interrupt Link [LMAC] enabled at IRQ 20
>> [ 1.249252] forcedeth 0000:00:0a.0: PCI INT A -> Link[LMAC] -> GSI 20 (level, low) -> IRQ 20
>> [ 1.249350] forcedeth 0000:00:0a.0: setting latency timer to 64
>> [ 1.302279] Loading iSCSI transport class v2.0-870.
>> [ 1.315354] forcedeth 0000:00:0a.0: ifname eth0, PHY OUI 0x732 @ 1, addr 90:fb:a6:2b:ee:ff
>> [ 1.315450] forcedeth 0000:00:0a.0: highdma csum pwrctl gbit lnktim msi desc-v3
>> [ 1.325446] ahci ...
>> [ 1.327784] scsi0 : ahci...
>> [ 1.330959] ata1: SATA max UDMA/133 ...
>> [ 1.352040] usb 2-3: ...
>> [ 1.648030] ata2: SATA link down (SStatus 0 SControl 300)
>> [ 1.648131] ata1: SATA link down (SStatus 0 SControl 300)
>> [ 1.652029] ata4: SATA link down (SStatus 0 SControl 300)
>> [ 1.652122] ata5: SATA link down (SStatus 0 SControl 300)
>> [ 1.652132] ata3: SATA link down (SStatus 0 SControl 300)
>> [ 1.656026] ata6: SATA link down (SStatus 0 SControl 300)
>> [ 1.672896] iscsi: registered transport (tcp)
>> [ 1.683754] iBFT detected at 0x93350.
>>
>>
>> _______________________________________________
>> gPXE mailing list
>> gPXE at etherboot.org
>> http://etherboot.org/mailman/listinfo/gpxe
>>
>>
>
More information about the gPXE
mailing list