Re: [cobalt-users] OS 3.0 Update Nightmare

[Jan Tietze <jptietze@xxxxxxxxxxx>]

Actually, reading your post and the corresponding error message makes me think
more of faulty network hardware and/or corrupted memory in the RaQ, resulting in
corrupted commands in shell scripts. You should probably extract the package
manually just to find out (the procedure to do this was on this list some days
ago IIRC) if there really *is* the string "echp" in a shell script inside the
kernel-source RPM. Kernel OOPS messages should never occur during an update
IMHO.

Since you commented on the RaQ philosophy itself, I took the freedom to comment
on a few aspects of your post.

As far as I am concerned, yes, there is room for improvements in the RaQ,
especially regarding extending the GUI (why is there no mechanism for extending
the GUI in a sane fashion?), but the overall product is excellent; it is far
better in many respects than plain linux boxen (of whom I have quite a few
running) for the specific purpose - a "no worries", extensible web hosting
solution. Its design is far from ideal  for people with very limited linux/*nix
clues, but to me and many others, this can also be regarded as an advantage over
truly dedicated web servers (of which few exist): ultimate extensibility in
terms of software. I can have PHP, MySQL, and many less-popular packages
installed that I would be missing badly had Cobalt decided not to build an open
platform. The RaQ3 really is a Red Hat 6.0-based open platform with
standard-grade hardware in a 1U box. Why not buy an Intel 1U box then instead?
Because cross-management (meaning: administer multiple services or aspects of a
service in the same UI, oftentimes even on the same page, avoiding repetitive
tasks like restarting daemons etc.) is a Good Thing (tm); developing your own
suite of programs to do the same takes time (buying a RaQ is cheaper), many
frequently needed services on the RaQ come pre-configured, and you need not
worry about individual package updates. Plus, it's a standard release when you
buy it, you'll get the same if you buy another. This is why I chose buying a
RaQ3 this time instead of building another Red Hat-or-whatever based box.

Jan

David Hall wrote:

> Well I have tried installing the OS 3 update 3 times
> now, and each time it has crashed the server and required a hard reboot.
> I tried installing it from the shell, GUI and a serial connection. No luck
> You know Cobalt is reminding more and more of Microsoft with all
> these patches, I bought this machine to make hosting set ups quicker and
> simpler
> but it has been just the opposite, it takes more time to admin these boxes
> than a straight Linux box,
>  I am not impressed and I think they let this box out the door too soon.
> Here is what I got, any help would be appreciated.
>
> /usr/local/sbin/cobalt_upgrade RaQ3-en-OSUpdate-3.0.pkg
> 4015 Problem installing package component: kernel-source-2.2.14C5-1.i386.rpm
> /tmp/upgrade_dir/upgrade_me: echp: command not found
> 4015 Error in installation script upgrade_me
> [root@xxxxx packages]# Unable to handle kernel NULL pointer dereference at
> virtu
> al address 00000074
> current->tss.cr3 = 00101000, %cr3 = 00101000
> *pde = 00000000
> Oops: 0002
> CPU:    0
> EIP:    0010:[<c0125a11>]
> EFLAGS: 00010002
> eax: 00000040   ebx: 00000001   ecx: c1a02540   edx: c7f65bd0
> esi: c1a02540   edi: 00000002   ebp: c7fe5f94   esp: c7fe5f44
> ds: 0018   es: 0018   ss: 0018
> Process kupdate (pid: 3, process nr: 3, stackpage=c7fe5000)
> Stack: c1a02540 c0126049 c1a02540 0000005c 00000002 c0181292 c0181177
> c1a02540
>        00000000 c7fe5fdc 00000003 00000080 00000002 03010001 0000000c
> 00000246
>        00000080 00000055 00000000 00032bea 00000001 c0181441 00000003
> 00000001
> Call Trace: [<c0126049>] [<c0181292>] [<c0181177>] [<c0181441>] [<c0126f5f>]
> [<c
> 01c812b>] [<c0127360>]
>        [<c0106000>] [<c010855b>]