Can't get a block out edgewise

From: Jeffrey E. Hundstad (
Date: Thu Aug 03 2000 - 19:43:15 EST


Here's the setup:

I have a Pentium-II 233MHz-128MB RAM, with a DPT RAID card with 64 MB
cache. The 17G of space is a RAID 5 set of 3-9G 10,000 SCSI drives.
I've got this mounted as an EXT2 filesystem, running on 2.2.17pre15 (but
it's the same for 2.2.15pre19.) Disk report below. Bonnie reports the
speed as:

              -------Sequential Output-------- ---Sequential Input--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block---
          MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
/sec %CPU
          100 851 23.8 615 1.1 330 1.1 1436 37.6 4871 4.2
559.4 4.6

Slow... Yea, I know...

Here's the problem:

I open two sessions. In one I say "find /", in the other I say "dd
if=/dev/zero of=big.file bs=10240K count=3".

When I get enter on the second sessions the first session STOPS
COMPLETELY until all the buffers from the second session completely

Here's the question:

Is there any way to smooth out the performance a bit? I'd like some
writes to occur then some reads so file I/O doesn't grind to a halt.

I've played around with /proc/sys/vm/bdflush and I can't seem to make
anything better than the standard behavior.

dmesg (disk report):
EATA/DMA 2.0x: Copyright (C) 1994-1999 Dario Ballabio.
EATA config options -> tc:y, lc:y, mq:62, eh:y, rs:y, et:n.
EATA0: 2.0C, PCI 0xff90, IRQ 11, BMST, SG 122, MB 64.
EATA0: wide SCSI support enabled, max_id 16, max_lun 8.
EATA0: SCSI channel 0 enabled, host target ID 7.
scsi0 : EATA/DMA 2.0x rev. 5.11.01
scsi : 1 host.
  Vendor: DPT Model: RAID-5 Rev: 07M1
  Type: Direct-Access ANSI SCSI revision: 02
Detected scsi disk sda at scsi0, channel 0, id 0, lun 0
EATA0: scsi0, channel 0, id 0, lun 0, cmds/lun 62, sorted, tagged.
scsi : detected 1 SCSI disk total.
SCSI device sda: hdwr sector= 512 bytes. Sectors= 35565312 [17365 MB]
[17.4 GB]

Thanks in advance for any clues.


Jeffrey Hundstad Minnesota State University, Mankato Mankato, Minnesota USA for pgp/gpg keys

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to Please read the FAQ at

This archive was generated by hypermail 2b29 : Mon Aug 07 2000 - 21:00:12 EST