I think I've started to localize my problem:
(for illustration, output from "cdrecord -scanbus"):
scsibus0:
0,0,0 0) *
0,1,0 1) 'YAMAHA ' 'CRW8424S ' '1.0j' Removable CD-ROM
0,2,0 2) 'IBM-PSG ' 'DPSS-318350N F ' 'S94S' Disk
0,3,0 3) 'HP ' '4.26GB A 50-0430' '0430' Disk
0,4,0 4) 'HP ' '4.26GB A 50-0430' '0430' Disk
0,5,0 5) *
0,6,0 6) *
0,7,0 7) *
The two HP disks, devices 3 and 4 share the 8-bit cable with the CD burner
(device 1), but more importantly, I've configured them as software RAID-0
devices (striping), using the newer (a.k.a. "Mingo") raid patches to my
2.2.18 kernel. The IBM disk, device 2, is on the U2W cable.
When I tried to burn an ISO that is stored on the Raid array, the burn
failed almost consistently, regardless of the speed or media-type. (I even
tried Kodak Gold Ultimas, to fully rule out poor media).
When I tried to burn the same ISO which had been moved to the IBM disk, the
burn was consistently successful (incidentally, on crappy CMC media...).
This would appear to suggest that either:
1. The software Raid-0 kernel patch overhead (i/o or cpu) is bogging down
things enough to cause the problems,
2. Reading the ISO from disk(s) on the same 8-bit cable as the burner is
causing underruns, or
3. The raid array is simply too slow to prevent underruns, period -
regardless of which which its on.
I would tend to rule out #3 since 512Mb 'bonnie' benchmark tests indicate
that the raid array, while considerably slower than the U2W IBM drive, still
puts in respectable throughput results. I've shown some bonnie results
below:
2000/12/21 -------Sequential Output-------- ---Sequential
Input--- --Random--
RH7 2.2.18-CHDS -Per Char- --Block--- -Rewrite-- -Per
Char- ---Block--- --Seeks---
biblios MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec
%CPU /sec %CPU
/dev/sda1 (/) 512 2876 99.2 28838 75.3 11333 51.3 2902 98.7 29839
67.2 421.1 9.2
/dev/sda4 (/home) 512 2873 99.1 26359 68.0 10785 48.1 2895 98.5 27725
60.2 417.1 11.7
/dev/sda5 (/usr) 512 2863 99.0 27221 72.2 10884 48.8 2898 98.6 28217
61.2 410.3 10.2
/dev/sda2 (/in) 512 2868 98.4 28886 69.4 7042 33.2 2917 99.2 29240
53.7 421.5 11.3
/dev/md0 (/storage) 512 2879 98.4 17760 34.9 5313 22.8 2900 98.7 17100
29.2 261.1 6.
I have trouble believing that #2 is the problem, since this is a common, if
not typical, configuration.
Unless I've missed something, then the issue would appear somehow relate to
use of the software Raid kernel patch itself. I'll try to confirm this by
backing up the data on the Raid array, breaking it up, and then repeating my
burn tests one one of the single HP disks on the same 8-bit cable. If this
turns out to be the case then I'm not sure what the problem could be - IO
contention, CPU contention, generally bad patch coding, etc.
I've searched through earlier messages on the cdwrite mailing list, but I
could find nothing posted regarding the combination of software raid-0
arrays with CD-burning. Does anyone have any experience with this
(successful or otherwise)?
In any event, it's been quite educational for me - not to mention
frustrating. At least now I know how to consistently burn CDs!...
Thanks again for everyone's help!
Jim
-----Original Message-----
From: Joerg Schilling [mailto:[EMAIL PROTECTED]]
Sent: Saturday, February 03, 2001 7:17 PM
To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
Subject: Re: Unpredictable Failures Burning CD-R
>From: "James Nikolich" <[EMAIL PROTECTED]>
>First of all, thanks for all the help in the past - I initially thought
that
>my problem had been solved by replacing the SCSI cable, but it now appears
>that it hasn't. Should've done more burns.... ;-)
>Envrionment:
> AMD K6II w/128Mb
> RedHat Linux 7.0 upgraded to kernel 2.2.18
> Tekram 390U2W adapter
> Yamaha CDR8424 internal SCSI CDRom
> CDRecord 1.9
> Media: BASF CD-R Extra 700Mb
>Track 01: 40 of 649 MB written (fifo 100%).
>Track 01: 41 of 649 MB written (fifo 100%).
>Track 01: 42 of 649 MB written (fifo 100%).
>cdrecord: Input/output error. write_g1: scsi sendcmd: retryable error
>status: 0x2 (CHECK CONDITION)
>CDB: 2A 00 00 00 55 FA 00 00 1F 00
>Sense Bytes: 70 00 05 00 00 00 00 0A 00 00 00 00 21 00 00 00
>Sense Key: 0x5 Illegal Request, Segment 0
>Sense Code: 0x21 Qual 0x00 (logical block address out of range) Fru 0x0
>Sense flags: Blk 0 (not valid)
This is a buffer underrun, check your SCSI setup.
Jörg
EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353
Berlin
[EMAIL PROTECTED] (uni) If you don't have iso-8859-1
[EMAIL PROTECTED] (work) chars I am J"org Schilling
URL: http://www.fokus.gmd.de/usr/schilling
ftp://ftp.fokus.gmd.de/pub/unix
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
