Cobalt RaQ/Qube Systems -- Testing Needed

Message

don_thomaso · Post by **don_thomaso** » Sun Jul 18, 2004 8:10 am

I'm running 2.6.6.
Should i enable the preemptible kernel option? Would it make any difference for a router, web, ftp, mail and ssh server?

deephack1982 · Post by **deephack1982** » Tue Jul 20, 2004 7:12 pm

Hello all,

I didn't notice this thread until recently. I have a Qube 2 and have been wondering about installing Gentoo on it for some time. I have all my other machines running it across 3 different architectures and this is the only one I haven't tamed yet.

I haven't managed to get the serial console to work. It may just be my cable but I get a black screen when using xc. Even when I set bps 115200 and 8N1. Anyway just wondering what people think is the best way to setup one of these systems just now. Netboot or copying a drive image while the drive is in another machine?

Regards,

DingbatCA · Post by **DingbatCA** » Wed Jul 21, 2004 5:26 am

Hey deephack1982.
I just got Gentoo installed on a Cobalt Raq2, not quite a Qube but I can give you a my 1..2..3's.. They should be the same.
To get the Qube to switch to console, just hold the reset switch while booting (on power on), it should give you an output on the LCD console on. From there you should have serial that works. Just give it a quick check on your serial settings bps 115200 and 8N1 no flow control. That should give you consol access. From there follow Kumbas instructions. http://dev.gentoo.org/~kumba/mips/cobalt/netboot/README These instruction are for a cobalt RaQ2 but also work for a Qube2. Here is the base of Kumbas page http://dev.gentoo.org/~kumba/mips/cobalt/ That should get you rolling. The only other MAJOR hurdle I came across was that GCC was not compiling correctly. Kumba was able to give me a fix for that one. #gcc-config mipsel-unknown-linux-gnu-3.3.3
Hope that helps.
DingbatCA

DingbatCA · Post by **DingbatCA** » Wed Jul 21, 2004 6:15 am

So my Cobalt RaQ2 has some issues, mostly because of compilling with GCC 3.4... so i told GCC to use 3.3.3; then ran an "emerge --update world --emptytree" which should recompile the entire system, i think. But it has been running the "checking" part of make for the last 2 day's. Is it checking every package befor compiling?? Any ideas?
-DingbatCA

BadAndy · Post by **BadAndy** » Wed Jul 21, 2004 11:46 am

I have been having troubles getting GCC 3.4 to compile on my RAQ2 also. How did you specify to use the older GCC (3.3) when doing the update.

deephack1982 · Post by **deephack1982** » Wed Jul 21, 2004 11:52 am

Thanks for the reply dingbat, I tried slaving a drive to my qube and installing the stage 3 tarball from the original cobalt install and then chrooting across but I got errors about excecutable formats so I guess the cobalt kernel is too old. I'm moving on to the netboot method now.

The Qube's don't have reset buttons but I'll try holding the buttons I do have till I find the right one.

Have fun!

deephack1982 · Post by **deephack1982** » Wed Jul 21, 2004 11:54 am

Okay I feel silly, the reset will be that small recessed one next to the power then

DingbatCA · Post by **DingbatCA** » Wed Jul 21, 2004 3:07 pm

-BadAndy
Not sure if youre asking how to tell GCC 3.4 to act like 3.3 or if your system is not compiling 3.4 at all. So here is both fix's.

1. Installing GCC 3.3. Go to your portage GCC dir. "cd /usr/portage/sys-devel/gcc" in there is a list of all versions of GCC that you should be able to compile... something on the lines of "emerge gcc-3.3" A little side note, if you are going from stage 1 you system will try to compile GCC 3.4 or newest. use the "emerge -i gcc" to inject a "stub" that will make your system think it has the newest version of GCC installed, thats if you want to stick with an old compile of GCC.

2. GCC 3.X is installed and you want to force it to run as GCC 3.3.3. That one is a simple command, but I had to ask to find it. "gcc-config mipsel-unknown-linux-gnu-3.3.3"
Hope this helps.
-DingbatCA

deephack1982 · Post by **deephack1982** » Wed Jul 21, 2004 7:17 pm

Phew, okay been working on this one a bit but I'm still having some problems.

I had a straight through serial cable instead of a null modem one so after replacing that I have a working serial console now. My current problem is that the netboot instructions say

2. Download the file "cobalt-nfsboot-colo_<ver>.gz" and unpack it into a directory which can be setup as an NFS
export.

When I try to unpack it gzip says it's not in gzip format. So I try it as is. So when I netboot the Qube I get the following messge on the serial console

Cobalt Microserver Diagnostics - 'We serve it, you surf it'
Built Tue Dec 15 04:11:42 PST 1998

1.LCD Test................................PASS
2.Controller Test.........................PASS
5.Bank 0:.................................64M
6.Bank 1:.................................0M
7.Bank 2:.................................64M
8.Bank 3:.................................0M
9.Serial Test.............................PASS
10.PCI Expansion Slot....................**EMPTY**
12.IDE Test................................PASS
13.Ethernet Test...........................PASS
16.RTC Test................................PASS
Decompressing done
Jump_to_Real_Kernel: disk error, trying BFD again
Decompressing done

And there it stops.

Anyone seen this one?

-- extra --
I just noticed this in my syslog

Jul 21 20:26:51 dragon rpc.mountd: refused mount request from qube for /nfsroot (/): no export entry

Here is my /etc/exports

/nfsroot qube(rw,async,no_root_squash)

DingbatCA · Post by **DingbatCA** » Wed Jul 21, 2004 7:41 pm

Humm it looks like my system is looping when I "emerge --update world --emptytree". It goes through its checks then states "/bin/sh config.status --recheck" and starts over. it just keeps doing this. Here is the first bit of screen out put. I looks like its stuck on patch-2.5.9. It will just keep checking and never stop looping. Any ideas?
-DingbatCA

"bash-2.05b# emerge --update world --emptytree
Calculating world dependencies ...done!
>>> emerge (1 of 96) sys-devel/patch-2.5.9 to"

DingbatCA · Post by **DingbatCA** » Wed Jul 21, 2004 7:53 pm

-deephack1982
Try these. Both a copy of my nfsroot and my tftpboot dir.
http://67.160.238.197/nfsroot.tar.gz
http://67.160.238.197/tftpboot.tar.gz
and for my exports....
/nfsroot 192.168.1.0/255.255.255.0 (ro,insecure,sync,insecure_locks,no_root_squash)

Try those...
-DingbatCA

deephack1982 · Post by **deephack1982** » Wed Jul 21, 2004 10:15 pm

Thats me in business now! Thanks for your help dingbat. Dunno what I was doing wrong but your images worked. I took a while trying to get tftp working then I realised that it should be service tftp not tftpd in xinetd

Thank you very much

DingbatCA · Post by **DingbatCA** » Thu Jul 22, 2004 2:42 am

Ok... I finly got my cobalt RaQ2 running under its own power.
But the boot is not clean... Take a look at this and please comment if you have any idea's these errors. (Below) Also after the "Strating local... [ok]" I loose console output and i am forces to get in via ssh.
Any ideas
-DingbatCA

Couldnt open /dev/vc/8
Couldnt open /dev/vc/9
Couldnt open /dev/vc/10
Couldnt open /dev/vc/11
* Failed to set user font
[ !! ]

* Loading key mappings...Couldnt get a file descriptor referring to the console

* Error loading key mappings
[ !! ]

don_thomaso · Post by **don_thomaso** » Thu Jul 22, 2004 7:24 am

DingbatCA wrote:Ok... I finly got my cobalt RaQ2 running under its own power.
But the boot is not clean... Take a look at this and please comment if you have any idea's these errors. (Below) Also after the "Strating local... [ok]" I loose console output and i am forces to get in via ssh.
Any ideas
-DingbatCA

Couldnt open /dev/vc/8
Couldnt open /dev/vc/9
Couldnt open /dev/vc/10
Couldnt open /dev/vc/11
* Failed to set user font
[ !! ]

* Loading key mappings...Couldnt get a file descriptor referring to the console

* Error loading key mappings
[ !! ]

in /etc/inittab
try fixing this line:
# SERIAL CONSOLE
c0

respawn:/sbin/agetty 115200 ttyS0 vt100

then the console output should keep working.

DingbatCA · Post by **DingbatCA** » Thu Jul 22, 2004 3:14 pm

-don_thomaso
Thx for the console fix.
If any one has any ideas on the (below) please pipe up.

Couldnt open /dev/vc/8
Couldnt open /dev/vc/9
Couldnt open /dev/vc/10
Couldnt open /dev/vc/11
* Failed to set user font
[ !! ]

* Loading key mappings...Couldnt get a file descriptor referring to the console

* Error loading key mappings
[ !! ]

mernisse · Post by **mernisse** » Thu Jul 22, 2004 7:46 pm

DingbatCA wrote:-don_thomaso
Thx for the console fix.
If any one has any ideas on the (below) please pipe up.

Couldnt open /dev/vc/8
Couldnt open /dev/vc/9
Couldnt open /dev/vc/10
Couldnt open /dev/vc/11
* Failed to set user font
[ !! ]

* Loading key mappings...Couldnt get a file descriptor referring to the console

* Error loading key mappings
[ !! ]

use rc-update to remove the font and keymap scripts from boot time :)
make sure your inittab isn't trying to start gettys on those devices either, you should only have the serial getty.

DingbatCA · Post by **DingbatCA** » Fri Jul 23, 2004 6:17 am

-mernisse
I dont mind just looking the other way on the /dev/vs stuff, that works :-)

ok, here is the newest one..

" * Checking root filesystem.../dev/hda3 has gone 49710 days without being checked, check forced."

And my clock is correct. this happes every time. its driving me nutz and its a 120GB disk!!! Ideas??

Redhatter · Post by **Redhatter** » Fri Jul 23, 2004 8:16 am

DingbatCA:

Are you using EXT2 per chance? I'd strongly recommend looking at either EXT3 or ReiserFS -- both are quite reliable on this platform. And both are journalled.

In the meantime, I spose you could use tune2fs to adjust when automatic checks are done (have a look at the manpage), but otherwise, no idea.

DingbatCA · Post by **DingbatCA** » Fri Jul 23, 2004 3:50 pm

-Redhatter
I have the system set up with ext3 for the root disk, only /boot is ext2. At first I was using ReiserFS but I made the mistake of not compiling that into the kernel... DOE!! so i ended up pulling my disk out, copying all my data off, reformating in ext3, then copying all my data back. It was a big pain, but it saved me from having to reinstall. Once Colo was installed I could no longer boot off the network, so there was no easy way to recompile my kernel.

So is it possible to switch to ReiserFS without pulling out my disk and copying all my data off?? If it is possible to switch, would you be so kind as to give me the steps? (ReiserFS support is now in my kernel now)

DingbatCA · Post by **DingbatCA** » Fri Jul 23, 2004 7:28 pm

does any one have a kernel config for 2.6.X that is working on a cobalt raq 2 or a qube 2? If so would you mind posting it?
thx

deephack1982 · Post by **deephack1982** » Sun Jul 25, 2004 1:30 pm

Hi Dingbat,

Yeah I got this one to compile on my Qube 2. Hope it does the trick. Just go into kernel hacking and switch the are you using crosscompiler option on or off depending on your setup.

http://dev.gentoo.org/~kumba/mips/cobal ... 606.config
Have fun!

BadAndy · Post by **BadAndy** » Mon Jul 26, 2004 1:18 pm

-DingbatCA

" * Checking root filesystem.../dev/hda3 has gone 49710 days without being checked, check forced."

I am getting this behavior too. Have you been able to figure out why this is happening.

DingbatCA · Post by **DingbatCA** » Tue Jul 27, 2004 5:39 am

-BadAndy
Well I got a fix for the EXT3 check disk in far to many days problem. :-)

/dev/hda3 / reiserfs noatime 0 0

I took the long path... copy my system onto a different disk. That worked but it is what caused my "Couldnt open /dev/vc/0" issues; I still have not found a fix for that one.

-All
Can any one get better then about 2200kbytes/ sec on your cobalt raq2 or qube2? I am copying files from my windows boxs to my cobalt with the aid of samba. Perhaps smb.conf tweaks??

mernisse · Post by **mernisse** » Tue Jul 27, 2004 6:30 am

BadAndy wrote:-DingbatCA

" * Checking root filesystem.../dev/hda3 has gone 49710 days without being checked, check forced."
I am getting this behavior too. Have you been able to figure out why this is happening.

This is because the netboot kernel doesn't read the onboard RTC of the cobalt Qube/RaQ boards so the first time you boot it the last checked time is in like 1999...

Once it checked on mine with the new kernel and the corrected (thankyou ntpdate) RTC it stopped complaining.

DingbatCA · Post by **DingbatCA** » Tue Jul 27, 2004 8:18 pm

Question?
Does any one no if the Cobalt RaQ2 is a 64-bit CPU??
Here is what I no so far.

system type : Cobalt RaQ2
processor : 0
cpu model : Nevada V10.0 FPU V10.0
BogoMIPS : 248.83
wait instruction : yes
microsecond timers : yes
tlb_entries : 48
extra interrupt vector : yes
hardware watchpoint : no
VCED exceptions : not available
VCEI exceptions : not available

Cobalt RaQ/Qube Systems -- Testing Needed

New to Qube

Installing gentoo on a Qube

Humm, its never done that befor.

GCC 3.4 Problems

Systeming looping.

Woohoo

Cobalt RaQ2 up and running.

Re: Cobalt RaQ2 up and running.

ext3 check fix