performace problem, D state

performace problem, D state

[Petr Sebor]

Hello,

I am having problems with system performace and I am unable to find the 
real cause of it. I have already
asked some people and polluted several mailing lists, but unfortunately, 
I don't know much more than
on the beginning.

I am having dual Opteron sytem (with two Opteron 244s) on a MSI 
Master2-Far board and 1G of ram.
The system is debian unstable/sid with vanilla 2.6.x kernels. The CPUs 
are running in non-native
mode - the entire system is 32 bit only and I am on 2.6.8-rc3 kernel 
right now.

The board is built around VIA chipset, has two SATA disks connected and 
onboard TG3 gigabit
ethernet interface.

The problem: the system goes to crawl when there is bigger disk 
activity. Bigger disk activity equals to doing
'grep -r foobar *' on uncached kernel source or doing subversion 'svn 
update' on 10Gig database.

I have naively blamed the disk (sata) subsystem, but I have been told 
that the sata is not very likely the culprit.

While doing the 'svn update', grepping the source, or compressing file 
archives, I can usually see following in 'top':

1334 petr      18   0 20724  14m 7516 D  1.6  1.5   0:01.17 svn
153 root       5 -10     0    0    0 D  0.0  0.0   0:01.09 reiserfs/0

with both CPUs usually spending >90% in 'wait state'. At this time, the 
system becomes very non-interactive for
any other task.

The disks are using reiserfs filesystem, but I have also tried to create 
ext2 partition with the same results.
'heavy' disk activity brings the system to a crawl.

The hardware in question is:

0000:00:01.0 PCI bridge: VIA Technologies, Inc. VT8237 PCI bridge 
[K8T800 South]
0000:00:0f.0 RAID bus controller: VIA Technologies, Inc. VIA VT6420 SATA 
RAID Controller (rev 80)

Dmesg ouput does not contain anything interesting (at least for a person 
with little or no knowledge, like me)

Just for the reference, this is the 'sata' part.

libata version 1.02 loaded.
sata_via version 0.20
sata_via(0000:00:0f.0): routed to hard irq line 11
ata1: SATA max UDMA/133 cmd 0xC000 ctl 0xC402 bmdma 0xD000 irq 11
ata2: SATA max UDMA/133 cmd 0xC800 ctl 0xCC02 bmdma 0xD008 irq 11
ata1: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 
88:203f
ata1: dev 0 ATA, max UDMA/100, 488397168 sectors: lba48
ata1: dev 0 configured for UDMA/100
scsi0 : sata_via
ata2: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 
88:407f
ata2: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48
ata2: dev 0 configured for UDMA/133
scsi1 : sata_via
Using anticipatory io scheduler
  Vendor: ATA       Model: WDC WD2500JD-00F  Rev: 02.0
  Type:   Direct-Access                      ANSI SCSI revision: 05
  Vendor: ATA       Model: WDC WD360GD-00FN  Rev: 35.0
  Type:   Direct-Access                      ANSI SCSI revision: 05

I am really clueless at this point.... If you have any ideas, please let 
me know.

Thanks,
Petr

performace problem, D state

[Wojtek]

Hello,

I seem to suffer from the same problem as described by Petr
in this post: http://lkml.org/lkml/2004/8/10/126

Performance of my machine gets really ugly
when doing I/O. Untaring stuff or copying files
brings the machine down, responsiveness
is really poor and load average sometimes goes 
as high as ten (CPU usage is very low then,
but I got usually near 100% value in wa column in top).

This is K8T Neo MSI motherboard with VIA chipset, Athlon64 CPU
(one, it's not SMP mobo). I'm using only IDE controller (SATA disabled). 

lspci shows this IDE hardware:

0000:00:01.0 PCI bridge: VIA Technologies, Inc. VT8237 PCI bridge
[K8T800 South] (prog-if 00 [Normal decode])
0000:00:0f.0 IDE interface: VIA Technologies, Inc.
VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
(prog-if 8a [Master SecP PriP])

Already tried playing with apic/acpi different kernel boot-parameters, they
don't help much. I've also been told that changing elevator to
'deadline' doesn't
affect this behaviour.

I have seen also other people (IRC/linux forums) having this issue
with MSI motherboards
with VIA IDE chipset and amd64 architecture.

I'd appreciate any hints which could help us to locate
the cause of this behaviour.

-- 
wr.

performace problem, D state

[Wojtek]

Hello,

I seem to suffer from the same problem as described by Petr
in this post: http://lkml.org/lkml/2004/8/10/126

Performance of my machine gets really ugly
when doing I/O. Untaring stuff or copying files
brings the machine down, responsiveness
is really poor and load average sometimes goes 
as high as ten (CPU usage is very low then,
but I got usually near 100% value in wa column in top).

This is K8T Neo MSI motherboard with VIA chipset, Athlon64 CPU
(one, it's not SMP mobo). I'm using only IDE controller (SATA disabled). 

lspci shows this IDE hardware:

0000:00:01.0 PCI bridge: VIA Technologies, Inc. VT8237 PCI bridge
[K8T800 South] (prog-if 00 [Normal decode])
0000:00:0f.0 IDE interface: VIA Technologies, Inc.
VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
(prog-if 8a [Master SecP PriP])

Already tried playing with apic/acpi different kernel boot-parameters, they
don't help much. I've also been told that changing elevator to
'deadline' doesn't
affect this behaviour.

I have seen also other people (IRC/linux forums) having this issue
with MSI motherboards
with VIA IDE chipset and amd64 architecture.

I'd appreciate any hints which could help us to locate
the cause of this behaviour.

-- 
wr.