Bug 6829

Summary: LSI Logic Megaraid SAS not supported with install dvd
Product: Mageia Reporter: Richard Couture <rrc>
Component: InstallerAssignee: Thierry Vignaud <thierry.vignaud>
Status: RESOLVED WORKSFORME QA Contact:
Severity: enhancement    
Priority: Normal CC: mageia, mageia, thierry.vignaud, tmb
Version: CauldronKeywords: NEEDINFO
Target Milestone: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Source RPM: CVE:
Status comment:
Attachments: dmesg from Fedora 17 install disk
dmesg from MGA2 install disk
lspci -nvvvv from Fedora 17 install disk
lspcidrake -v from MGA2 install disk
/proc/bus/pci/devices from Fedora 17 install disk
/prov/bus/pci/devices from MGA2 install disk
dmesg diff

Description Richard Couture 2012-07-19 20:36:42 CEST
I want to install Mageia on a clients IBM x3500 M4 server which uses the
LSI Logic Megaraid SAS controller as do Dell servers and many others.

I am advised that there is no disc space available for the install.

SuSE 12, Fedora 17 and RedHat 6 all can see and install, so I know that
the HW is properly configured and functional

As near as I can figure, the initrd of the install doesn't have a
current megaraid-sas.ko.xz module.

Is this something the we might be interested in providing? or do I have
to learn how to create my own initrd and re-burn the distrib with it?

Thanks




Richard
Comment 1 Manuel Hiebel 2012-07-21 02:39:25 CEST
setting version to cauldron as it's an enhancement bug

CC: (none) => thierry.vignaud, tmb
Version: 2 => Cauldron

Comment 2 Thierry Vignaud 2012-07-23 14:22:08 CEST
Can you attach the lspci.txt & proc_pci.txt files resulting from running the following commands on any Linux distribution:

lspci -nvvvv > lspci.txt
cat /proc/bus/pci/devices > proc_pci.txt

Keywords: (none) => NEEDINFO

Thierry Vignaud 2012-07-23 14:22:18 CEST

Assignee: bugsquad => thierry.vignaud

Comment 3 Thierry Vignaud 2012-07-23 14:26:35 CEST
Also, as we do have megaraid_sas.ko.xz in the installer, we would need for you to attach screenshots (camera) of other consoles (just type alt+f3 & alt+f4 in order to access the installer screen as well as the kernel one).
Comment 4 Richard Couture 2012-08-31 19:23:07 CEST
Created attachment 2713 [details]
dmesg from Fedora 17 install disk
Comment 5 Richard Couture 2012-08-31 19:24:32 CEST
Created attachment 2714 [details]
dmesg from MGA2 install disk
Comment 6 Richard Couture 2012-08-31 19:25:28 CEST
Created attachment 2715 [details]
lspci -nvvvv from Fedora 17 install disk
Comment 7 Richard Couture 2012-08-31 19:26:13 CEST
Created attachment 2716 [details]
lspcidrake -v from MGA2 install disk
Comment 8 Richard Couture 2012-08-31 19:27:00 CEST
Created attachment 2717 [details]
/proc/bus/pci/devices from Fedora 17 install disk
Comment 9 Richard Couture 2012-08-31 19:27:49 CEST
Created attachment 2718 [details]
/prov/bus/pci/devices from MGA2 install disk
Comment 10 Richard Couture 2012-08-31 19:29:37 CEST
Sorry it took SSSOOOOooooo long. The customer could not find time to let me play with the server.  I now have the server in my office so as to preclude any further delays as I am quite eager to resolve this issue and install MGA2 without having to install it on a SLOWWWWW SATA disk.

I am attaching 6 files:
  lscpi-mga2.txt created from MGA2 install disk: lspci -nvvvv
  lscpi-mga2.txt created from Fedora17 install disk: lspci -nvvvv

  proc_pci-mga2.txt created from MGA2 install disk: cat /proc/bus/pci/devices
  proc_pci-fed17.txt created from Fedora17 install disk: cat /proc/bus/pci/devices

  dmesg-mga2.txt created from MGA2 install disk: dmesg
  dmesg-fed17.txt created from Fedora17 install disk: dmesg

I am at your disposal for any further requirements which you may have


Richard
Thierry Vignaud 2012-09-03 14:52:04 CEST

Attachment 2716 description: lspci -nvvvv from MGA2 install disk => lspcidrake -v from MGA2 install disk

Thierry Vignaud 2012-09-03 15:08:10 CEST

CC: (none) => mageia

Thierry Vignaud 2012-09-03 15:08:17 CEST

CC: (none) => mageia

Comment 11 Thierry Vignaud 2012-09-03 15:09:13 CEST
Created attachment 2738 [details]
dmesg diff

We do load the megaraid device and it looks like at least rescue detect all disks.
We don't load as much modules as FC does but regarding the storate, we load enough.

Can you start installer with, run it until you got the error, then:
1) plug an USB key
2) run the command bug that will write a bug.gz file
3) attach that file to this BR
Comment 12 Thierry Vignaud 2012-09-03 15:10:58 CEST
I supposed lspcidrake came from rescue but it may be from regular installer according to what you said?
Still attach the file generated by the bug command please :-)
Comment 13 Richard Couture 2012-09-05 14:59:57 CEST
(In reply to comment #12)
> I supposed lspcidrake came from rescue but it may be from regular installer
> according to what you said?
> Still attach the file generated by the bug command please :-)

While working with the machine in order to comply with your request, lo and behold, the damned thing died. It's LSI Logic RAID controller bit the dust and the machine has been returned to IBM for repair, soooo:

It will be a few days before I get it back, and then we'll see if this bug report is founded upon faulty hardware.



Richard
Comment 14 Thierry Vignaud 2012-09-07 19:16:15 CEST
I tried on a IBM machine (with Mageia 3 alpha1).
It works fine once a virtual (RAID) device is created using the
controller BIOS.
If you do not create any virtual device in the BIOS, there's no disk to see...

Reopen this bug report if you see this error after having created a virtual device in the BIOS

Status: NEW => RESOLVED
Resolution: (none) => WORKSFORME

Comment 15 Richard Couture 2012-09-25 17:25:07 CEST
The server finally returned from IBM warranty service yesterday and it actually functioned normally for about 2 hours. During those 2 hours Magea 2 had no problems seeing the raid array that was presented to it by the LSI Logic MegaRaid controller.

The server promptly failed after about 2 hours with the same hardware failure that we saw originally, but

  Magea 2 has been exonerated of any wrong doings

I apologize for the false error report. It was a hardware problem that was difficult to see until the whole thing went belly-up

Thanks



Richard