Description of problem: Can't install Mageia 2 RC2 on HP Proliant ML 350. Installer don't reconize disk in RAID 5. Version-Release number of selected component (if applicable): How reproducible: Try install on HP Proliant ML 350 with disks in RAID 5. Steps to Reproduce: 1. 2. 3.
CC: (none) => pterjan, thierry.vignaud, tmbSummary: Can't install Mageia 2 RC2 on HP ProLiant ML 350 => Can't install Mageia 2 RC on HP ProLiant ML 350
I suppose this is hardware raid and does not matter.
Could it be one of the controllers that need nonfree firmware ? If so, try with boot-nonfree.iso
The driver should be cciss which does not seem to need any firmware
Any errors in Alt-F2/F3/F4 ? can you plug a usb stick, check dmesg that it got recognized and type: bug that should store installer logs on the usb stick, and you can then attach the file to this report
Created attachment 2272 [details] Alt+F1 screen
Created attachment 2273 [details] Alt+F3 screen
Created attachment 2274 [details] Alt+F4 screen
I tried with boot-nonfree.iso, also don't worked. See photos from Alt+F1/F3/F4 in atach..
Hi, This bug was filed against cauldron, but we do not have cauldron at the moment. Please report whether this bug is still valid for Mageia 2. Thanks :) Cheers, marja
Keywords: (none) => NEEDINFO
(In reply to comment #9) > Hi, > > This bug was filed against cauldron, but we do not have cauldron at the moment. > > Please report whether this bug is still valid for Mageia 2. > > Thanks :) > > Cheers, > marja Yes. I can´t install Mageia 2 on this server []´s --- Eduardo
Keywords: NEEDINFO => (none)CC: (none) => sander.lepik
the file report.bug was also missing
Summary: Can't install Mageia 2 RC on HP ProLiant ML 350 => Can't install Mageia 2 on HP ProLiant ML 350
It doesn't look like we detected a cciss managed controller. Can you attach "lspcidrake -v" output from eg the rescue system? eg the pci.txt file resulting from running "lspcidrake -v>pci.txt" that you can just put on some USB key
(In reply to comment #12) > It doesn't look like we detected a cciss managed controller. > > Can you attach "lspcidrake -v" output from eg the rescue system? > eg the pci.txt file resulting from running "lspcidrake -v>pci.txt" that you > can just put on some USB key Sorry, I can't mount USB key with RW access in rescue system. There is another way? Or you can tell me how I mount USB key with RW access?
What is your error? Is it a NTFS or a FAT formatted one? Can't you run sg like "mount /dev/sdX /mnt"? or "mount /dev/disk/by-label/WHATEVER_NAME /mnt"?
(In reply to comment #14) > What is your error? > Is it a NTFS or a FAT formatted one? > Can't you run sg like "mount /dev/sdX /mnt"? > or "mount /dev/disk/by-label/WHATEVER_NAME /mnt"? It is NTFS formatted. I tried severals syntax: mount /dev/sda /mnt/usb mount -t ntfs -o rw /dev/sda /mnt/usb etc. In all cases the pendrive is mounted in "read only" mode.
Is there anything relevant in dmesg output?
(In reply to comment #16) > Is there anything relevant in dmesg output? No. Nothing relevant... What you say if I try "lspcidrake -v > pci.txt" using live CD system?
Yes, that should be the same. Though I would like to fix the ntfs formatted usb key issue too.
Created attachment 2783 [details] dmesg output
Created attachment 2784 [details] lspcidrake -v output
(In reply to comment #18) > Yes, that should be the same. > Though I would like to fix the ntfs formatted usb key issue too. I am sending in attach. lspcidrake -v and dmesg outputs. (I tested "Mageia 3 alpha 1" and notice the some problem...)
These outputs (lspcidrake.out and dmesg.out) are from Live CD system. I have good news: I tried install from Live CD installler in this server and I all works well (recognized all hard disks)
Yes but real installer still suffers from this bug, isn't it? It looks like your discs are attached to the "Hewlett-Packard Company|Smart Array E200i" SAS Controller, not to the "LSI Logic / Symbios Logic|53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI" controller or to the, aren't they ? What's strange is that ldetect reports it should be managed by "hpsa" instead of "cciss" whereas: 1) 0x103c:0x3238 is only manager by cciss, not by hpsa both checked by modinfo on both modules and by running "/sbin/modprobe -R pci:v0000103Cd00003238sv0000103Csd00003211bc*sc*i*" 2) there's no hpsa hardcoded in ldetct* _THIS_ is probably the issue with the drakx installer. I guess in the livecd case that the right driver is loaded by udev. @Colin: is there a way to trigger the same behaviour with udev in installer stage2?
Status: NEW => ASSIGNEDCC: (none) => mageiaAssignee: bugsquad => thierry.vignaudSource RPM: (none) => ldetect, ldetect-lst
(In reply to comment #23) > Yes but real installer still suffers from this bug, isn't it? Yes. I can't install from another way instead livecd.
Can you attach the pci.txt file resulting from running "cat /proc/bus/pci/devices > pci.txt"?
Created attachment 2787 [details] cat /proc/bus/pci/devices output
(In reply to comment #25) > Can you attach the pci.txt file resulting from running "cat > /proc/bus/pci/devices > pci.txt"? pci.txt is in attach.
CC: (none) => mageia
What's the output of "cat /sys/bus/pci/devices/*/modalias" ? Can you confirm than manually running "modprobe cciss" in the console (alt+f2) of the traditionnal installer makes it work?
Created attachment 2791 [details] cat /sys/bus/pci/devices/*/modalias output
(In reply to comment #28) > What's the output of "cat /sys/bus/pci/devices/*/modalias" ? See modalias.txt in attach.. > Can you confirm than manually running "modprobe cciss" in the console (alt+f2) > of the traditionnal installer makes it work? Yes, I can. "modprobe cciss" makes it work.
And the winner is: /sbin/modprobe -R pci:v0000103Cd00003238sv0000103Csd00003211bc01sc04i00 hpsa cciss The issue is that for some devices, the first driver is the best, for others, it's the last. Or maybe should I skip "hpsa" in ldetect if not the latest...
one way that could be tested if it works is to boot with: modprobe.blacklist=hpsa added to the kernel command line... if it does help, we could add it to errata for Mga2
Created attachment 2792 [details] pick latest driver This patch makes ldetect pick latest driver. This fixes this case. However, in one case, on my machine, I have: module0 = 'pata_atiixp' module1 = 'ata_generic' module2 = 'pata_atiixp' module3 = 'pata_acpi' module4 = 'atiixp' module5 = 'ide_pci_generic' We may have to rethink the config files ordering.
(In reply to comment #32) > one way that could be tested if it works is to boot with: > > modprobe.blacklist=hpsa > > added to the kernel command line... > > if it does help, we could add it to errata for Mga2 Yes. "modprobe.blacklist=hpsa" in boot options is a good workaround and solve the problem.
This should probably be fixed in /lib/modules/$(uname -r)/modules.order instead of patching kmod. See http://lwn.net/Articles/260856/
Humm we already have it. But we don't use it in stage1. However kmod uses it, but only in depmod (like module-init-tools). Eventually kmod would need patching in order to use modules.order when solving a modalias.
According to upstream: "kmod is correct to only ever reference the modules.order file when calling depmod. Aliases are sorted once according to the list, and then modprobe should then just Do The Right Thing⢠when resolving the alias from the generated index. The bug here is that the sorting fails in depmod when modules are compressed." But s/ko$/ko.xz/ on modules.order, then calling depmod -a results in : $ sbin/modprobe -R pci:v0000103Cd00003238sv0000103Csd00003211bc01sc04i00 cciss hpsa Likewise, ldetect is OK. Thomas, can you fix modules.order in our kernel?
(In reply to comment #37) > According to upstream: > "kmod is correct to only ever reference the modules.order file when > calling depmod. Aliases are sorted once according to the list, and then > modprobe should then just Do The Right Thing⢠when resolving the alias > from the generated index. The bug here is that the sorting fails in > depmod when modules are compressed." > > But s/ko$/ko.xz/ on modules.order, then calling depmod -a results in : > $ sbin/modprobe -R pci:v0000103Cd00003238sv0000103Csd00003211bc01sc04i00 > cciss > hpsa > > Likewise, ldetect is OK. > > Thomas, can you fix modules.order in our kernel? I supposed you guys are working to fix this in Mageia 3 alpha 1...
Yes, mga2 installer is frozen. In fact we're doing this in Cauldron, thus for mga3a2. Thomas, I've fixed temporary in drakx-installer-images but lspcidrake will still report wrong on installed system
This may be tricky to solve with modules.order, cciss is in kernel/drivers/block and hpsa is in kernel/drivers/scsi, they do not use per-module sub-directories.
Well it does work when modules.order is read fine by depmod when sorting modules.alias*
José, can you try latest boot-nonfree.iso from cauldron? eg: boot-nonfree.iso 17-Sep-2012 20:56 25M http://distrib-coffee.ipsl.jussieu.fr/pub/linux/Mageia/distrib/cauldron/x86_64/install/images/boot-nonfree.iso
(In reply to comment #42) > José, can you try latest boot-nonfree.iso from cauldron? > eg: boot-nonfree.iso 17-Sep-2012 20:56 25M > http://distrib-coffee.ipsl.jussieu.fr/pub/linux/Mageia/distrib/cauldron/x86_64/install/images/boot-nonfree.iso Sorry. I tried this boot-nonfree.iso. It don't detect cciss automatically. Only work if i select by hand the driver.
Can you try boot-nonfree.iso from now that should land on distrib-coffee in an hour?
(In reply to comment #44) > Can you try boot-nonfree.iso from now that should land on distrib-coffee in an > hour? I can't install with this new boot-nonfree.iso. There is an error in X server. In mode text I get this another error: "insmod' ing module ide_core, ide_cd_mod, scsi_mod, sr_mod failed"
Can you open another bug report for your issue with the graphical mode? Can you attach the report.bug file (after compressing it of course) to this bug? You can get that file by plugging a USB key then running the "bug" command in the second tty (but after encountering your bug)? Did you see any error messages in tty3 or tty4 (such as "missing module ...")?
I don't have access to tty using Alt+F2. I supposed installer don't reached this step... In Alt+F3 I can see these messages: * starting step `setupSCSI´ * missing module ide_core * missing module ide_cd_mod * missing module scsi_mod * missing module sr_mod
Can you try again? Stage2 of the installer got broken for a couple hours due to latest dash having introduced a new symlink :-(
Good job, Thierry. It works. The installer recognized the hard disks. Notice that I can't pass after this step and install Mageia 3 alpha because I have Mageia 2 in this machine, installed with live CD. This system is our server backup.
Closing then.
Status: ASSIGNED => RESOLVEDResolution: (none) => FIXED
*** Bug 8280 has been marked as a duplicate of this bug. ***
CC: (none) => mrmazda