[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

DCG 533 Review, Howto configure, etc.

Purchasing finally got a PO to DCGINC on 3/13/98, I got the system today
not bad.  (A week and a day).

The PO had everything I wanted on it, so I go to plug it in.... NOTHING.

Go to take the case apart, theres no manual for it.  It SEEMED like the
front cover would come off, but it didn't.  Turns out you start with the
rear screws and both sides come off easily.  Nice case design, very well
made, no "fiddling" needed to reassemble.  Kinda loud though, only a single
4.5 gb 5400 rpm scsi inside.

Turns out the bracket for the speaker/fan had detached and bounced around
inside the case.  The speaker was punctured, the wires ripped off (they
were soldered), the fan came out of the housing.  I reattached the fan 
inside the housing and the housing to the case.  Not sure if I got
it upside down or not (it fits both ways).  The power cables were 
disconnected from the motherboard as well.  BTW if you accidently plug 
in a keyboard to the mouse part it won't boot at all.

Hooked up everything and it booted, got the the alphabios prompt, which
had a milo entry.  Unfortunately it was set to boot a: (the floppy).  DCG 
included the full redhat-5.0 distribution (a nice touch) complete with 
boot floppies (milo/linload, boot disk, and ramdisk).

I booted the milo floppy and found that actually redhat-5.0 had been installed
just didn't have milo configured.  I say at the MILO> prompt:
	 boot sda2:vmlinux.gz root=/dev/sda2 

The partitions were kinda strange, but I guess everyone has a preference:
   Device Boot   Begin    Start      End   Blocks   Id  System
/dev/sda1            1        1        1     4340    4  DOS 16-bit <32M
/dev/sda2            2        2       48   205437   83  Linux native
/dev/sda3           49       49      168   524520   82  Linux swap
/dev/sda4          169      169     1018  3715350    5  Extended
/dev/sda5          169      169      520  1538561   83  Linux native
/dev/sda6          521      521     1018  2176727   83  Linux native

Filesystem         1024-blocks  Used Available Capacity Mounted on
/dev/sda2             198910   69786   118853     37%   /
/dev/sda6            2105438    7600  1989002      0%   /home
/dev/sda5            1488740  620064   791748     44%   /usr

Looks like DCG or Redhat upgraded /usr/src/linux which causes alot of things 
to break, i.e. compiling egcs/ssh etc.  I replaced /usr/src/linux and things
compiled fine I.e. egcs-1.0.2 and ssh-1.2.22.

I downloaded all the redhat patches, and it turns out most of them
had already been installed (a nice touch).

Some benchmarks:
Function      Rate (MB/s)   RMS time     Min time     Max time
Copy:         219.0377       0.1837       0.1826       0.1875
Scale:        212.2274       0.1895       0.1885       0.1914
Add:          230.1126       0.2613       0.2607       0.2637
Triad:        214.8251       0.2804       0.2793       0.2832

Building the entire egcs-1.0.2 tree:
553.21user 46.28system 10:10.70elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (537031major+430748minor)pagefaults 0swaps

              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
         1024  5098 41.2  5497 26.2  2378 15.0  4950 35.9  5492  7.2  88.0  1.3

 Timing buffered disk reads:  16 MB in  2.38 seconds = 6.72 MB/sec

[root@freesurface /root]# cat /proc/cpuinfo | grep Bogo 	;-)
BogoMIPS                : 530.57	

Kinda dissapointing disk bandwidth (about 1/2 of a IBM EIDE+dell), and high
cpu usage (about double of IBM+Dell).  But this was purchased as a cpu
server not a file server, disk should not be a bottle neck.  I suspect the 
blame is the largely the low end quantum Viking disk.  

Oh if your recompiling the kernel make sure to pick the bsd version of the
NCR driver, I couldn't get the linux ncr driver to see the controller I.e. 
panic on mounting 08:02.  

Once I reformated /dev/sda1 in a msdos format (mkfs.msdos), copied over milo
and linload.exe from the DCG provided boot floppy, and tweaked milo I got
it to boot standalone.  In milo I set:

name:	linux
boot file:disk 0, partition 1	liload.exe
os path:  disk 0, partition 1	milo
os options: boot sda2:vmlinux.gz root=/dev/sda2

Theres also a alpha-bios setup option to auto-start and I set it at 30 seconds.

Oh btw do NOT assume /sbin/shutdown -r now syncs the disk.  I did a:
cp /usr/src/linux/alpha/boot/vmlinux.gz /;shutdown -r now

When it rebooted the files were different.  I'll add a sync in there from
now on.

My only remaining problem is amd isn't happy even after (from redhat errata):
rpm --force -Uvh --noscripts portmap-4.0-8.alpha.rpm glibc-2.0.7-4.alpha.rpm amd-920824upl102-11.alpha.rpm 

Anyone else have a suggestion on fixing amd/nfs?

After leaving it up overnight I get the following via dmesg:
ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)><sc 94(11ffffc90,1,ea60)>

Any ideas?

So in summary the machines up, reasonably happy, fast.  Still have some 
tweaking to do.

Bill Broadley           Bill@math.ucdavis.edu               UCD Math Sys-Admin
Linux is great.         http://math.ucdavis.edu/~bill			PGP-ok

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []