Results 1 to 6 of 6
Hi!
First, a little background: I'm not a "noob" about *nix. I currently run two Debian servers (that have been running since Debian 3.0, now at 5.0) my main computer ...
- 12-23-2010 #1Just Joined!
- Join Date
- Dec 2010
- Posts
- 3
Is it me/Fedora/hardware?
Hi!
First, a little background: I'm not a "noob" about *nix. I currently run two Debian servers (that have been running since Debian 3.0, now at 5.0) my main computer has, at various points, been running RedHat (back when it was called that) and Ubuntu. I'm currently in the process of installing a new server and a laptop, and these are the ones I have problems with.
First, the server. It is an old-ish Dell, Precision 650 (with only one CPU). SCSI has two tape-drives on it. There's a DVD-drive and a ~300Gb HDD connected to the primary IDE channel. I added a Sil3112-based PCI card with two ports to and connected two (Seagate and WD) 1TB SATA drives to it. Then I tried to install Fedora Core 14 (off a DVD).
That's when the problem started. I configured the ~300Gb IDE disk as a boot drive with FC default configuration (lvm) and made a lvm group and single volume out of the two SATA drives. The installer took its sweet time to format the disks, but I gathered it was because it is an older computer and the 2TB space is quite a lot. However, at first boot after installation it dropped down to disk maintenance mode claiming fs errors. I hit fsck -y and watched it fix errors for almost two hours. (All drives are new and empty so I'm not afraid of losing data or the like.) I though that it was weird having errors on empty drives after install, but went with it. Computer rebooted and voilá - back at the maintenance mode. Errors popping up left and right.
I blamed the lvm, reinstalled with simple partition scheme and made the two disks separate mount points with just single primary partition on both disks. installer aborted. Tried again and left sata drives unpartitioned - everything worked. Once up and running, fired up GParted and partitioned sata drives as single partitions - errored while formatting with fs errors.
Long-ish story shorter, I couldn't get the two SATA disks to work in ANY way I could figure out on Fedora. Was fed up with that so popped in Debian 5 netinstall CD, installed and once up and running configured a lvm like I first tried with FC - no errors. Everything seems to work fine at least. So, the question(s) is(are):
Do I just do something wrong or does Fedore Core 14 really support
a) Sil3112-based SATA controllers, b) 1TB sata drives?
Or doesn't my controller support two 1TB drives on it? (and how can I tell?)
And why does Debian then work - or, worst option - it doesn't but it just doesn't show the errors and I'm headed towards data corruption and disk failure. (how to test if the disks/lvm really works?)
Then, the laptop. I have a HP/Compaq 6715s laptop and tried to install FC14 on it. Worked othewise fine but can't for the life of me figure out how to make wireless work. Is that piece of cr*p laptop just not compatible, or what am I missing?
(Tried that same laptop with ubuntu and didn't even got picture -_-, debian netinstall needed eth0 so I couldn't test as I hadn't physical net access at that time.)
EDIT: Got that WiFi working on this laptop with a bit o' googling and RPM Fusion. Now running happily. (ZTE MF668+ 3G Modem worked by just inserting the thing and making a connection! w00t! No hassle with drivers or anything! happy!)Last edited by AnttiV; 12-23-2010 at 11:36 PM.
- 12-24-2010 #2
I have a couple generations earlier Precision 530 as my home server, and it won't let me use the IDE controller and my add-in PERC 3 RAID controller at the same time. You may be hitting a similar limitation on your 650. Or not.
For your Compaq wireless, it appears to have a Broadcomm 4300 in it. Getting this to work is a LOT easier than it used to be, when you had to use b43cutter to chop the firmware out of a Windows package. Now you can just add the RPMFusion free and non-free repositories at:
RPM Fusion - RPM Fusion
Then do
yum install broadcom-wl
Then create your connection in NetworkManager and it should just work.
Don't worry, "non-free" doesn't mean there's a cost, just that the stuff isn't open-source licensed.
Oh, just noticed your edit. Maybe this will help someone else with the details.
- 12-24-2010 #3Just Joined!
- Join Date
- Dec 2010
- Posts
- 3
That - or similar - was what I thought at first. Specifically, I thought the controller might not like the 1TB disk, since it's a bit older component. I don't know what the disk size really has anything to do with it, but I've come across such thing before. However, now as it (at least seems to) works fine with Debian, I'm not so sure. I mean, it still COULD be like that and Debian just fails to report the errors that FC did or that Debian doesn't scan/format the disks as FC does and thus fails to replicate the errors. (maybe Deb just does a "quick format" and doesn't really query the disks afterward at all? so no errors reported? At least deb's mkfs went a H*LL LOT faster than what FC managed.)
- 12-24-2010 #4
Have never been a Debian user, so I don't know what it may be doing differently. Looking back, I can see that there's been a lot of funkiness over time particularly with RH-like distros WRT the sata_si and sata_si24 drivers, things like device files changing with kernel updates and weirdness when used in conjunction with IDE/PATA controllers.
I'd try hitting the Debian LV with some heavy traffic, like rsyncing the IDE drive to it on a couple of different target directories. I'm betting it goes belly up, flips read-only or the like.
Re the format time, I just installed Centos 5.5 onto a 2Tb RAID 5 (hardware, one of the Intel LSI-based cards, uses megaraid driver) this week, with straight partitions for /boot (500Mb) and / (50Gb), and a ~1.9Tb LVM for /home. Total format time for the installer was about 75min. Physical disks are 3 1Tb 7200's. I'd guess that a non-RAID should be faster.
- 12-25-2010 #5Linux Engineer
- Join Date
- Feb 2005
- Posts
- 1,044
Many moons ago, I was involved with a VAX/780 that gave fsck errors which got worse the more we ran it. Turned out to be a bit stuck on in each word of 4MB of RAM that happened to be used to hold the bit map. Replacing the RAM "fixed" the disc.
- 12-25-2010 #6Just Joined!
- Join Date
- Dec 2010
- Posts
- 3
Thanks for the rsync idea, I'll try that. If that works ok for a few days, I think I'm good with that. I would like to use Fedora for the server though, but I have no idea how to make it work...
Ookay, that's a solution that probably wasn't the first you'd think about
Luckily, the RAM I have on the server is old, so I ran it through memtest in the first place to see if it would work. No errors there so I should be ok.
Honestly, I think it would take less resources to just buy a new comp and ditch this Dell
If it wasn't for personal use and if I wan't so short on money, I would


Reply With Quote
