(Solved) Kernel 2.6.35 I/O congestion

Onyros · 2010-08-25 12:31:07

The difference after downgrading is really dramatic The thing is... it happened on all my machines running Arch, and they vary a bit in terms of specs - only thing in common between them is that they all have Intel CPUs.

karol · 2010-08-25 12:48:28

Onyros wrote:

The difference after downgrading is really dramatic The thing is... it happened on all my machines running Arch, and they vary a bit in terms of specs - only thing in common between them is that they all have Intel CPUs.

Indeed, my Sempron box doesn't show any I/O congestion.

Bl@ster · 2010-08-25 13:20:48

karol wrote:

Onyros wrote:
The difference after downgrading is really dramatic The thing is... it happened on all my machines running Arch, and they vary a bit in terms of specs - only thing in common between them is that they all have Intel CPUs.
Indeed, my Sempron box doesn't show any I/O congestion.

Well, Sempron is AMD stuff

Anyone tried with kernel26-git?

Last edited by Bl@ster (2010-08-25 13:21:29)

karol · 2010-08-25 13:22:24

Bl@ster wrote:

karol wrote:
Onyros wrote:
The difference after downgrading is really dramatic The thing is... it happened on all my machines running Arch, and they vary a bit in terms of specs - only thing in common between them is that they all have Intel CPUs.
Indeed, my Sempron box doesn't show any I/O congestion.
Well, Sempron is AMD stuff

Yeah, I know. I don't boot it every day and I've been too busy to check if it's having I/O problems with the new kernel, but the hint from Onyros prompted me to do so.

Bl@ster · 2010-08-25 17:30:36

Any news?
So impatient

ffjia · 2010-08-26 02:58:16

Mine, after a few minutes of idle:

load average: 0.12, 0.27, 0.34

i5, 2g ram, 2.6.35-3, kde 4.5

It seems normal, right?

I want to know that you guys have had bad user experience using 2.6.35 besides
higher load average?

Bl@ster · 2010-08-26 07:19:18

ffjia wrote:

Mine, after a few minutes of idle:
load average: 0.12, 0.27, 0.34
i5, 2g ram, 2.6.35-3, kde 4.5
It seems normal, right?
I want to know that you guys have had bad user experience using 2.6.35 besides
higher load average?

My operations are the same of 2.6.34. Surf the internet, read my mail, chat... daily things
However, try to leave your machine: the load raises especially doing simple duties like dragging a window; you don't need to force your CPU, it happens when it wants

karol · 2010-08-26 10:44:52

ffjia wrote:

Mine, after a few minutes of idle:
load average: 0.12, 0.27, 0.34
i5, 2g ram, 2.6.35-3, kde 4.5
It seems normal, right?
I want to know that you guys have had bad user experience using 2.6.35 besides
higher load average?

https://bbs.archlinux.org/viewtopic.php … 34#p814034
I have no problems.

strummer · 2010-08-27 02:07:21

load average: 1.01, 1.27, 1.33 doing nothing, but Deluge is running.

I just upgraded and noticed that mplayer says it's too slow to play some files. Also when I open any web page with Chromium CPU usage goes to 100% for a couple of seconds.

Wintervenom · 2010-08-27 02:48:50

Using Zen 2.6.35 with BFS and BFQ schedulers on an AMD Athlon64 X2 TK-55, an Intel Atom N450, and a VIA Esther C7-D. No load problems so far.

incubine:~> iterate "ssh-cmd ::: uptime" incubine succubine cannabalis
<ssh-rcmd@incubine.incubine.demonpit>  21:48:25 up 20:50,  1 user,  load average: 0.20, 0.31, 0.21
<ssh-rcmd@succubine.incubine.demonpit>  21:48:26 up 59 min,  1 user,  load average: 0.11, 0.34, 0.18
<ssh-rcmd@cannabalis.incubine.demonpit>  21:48:26 up  3:59,  0 users,  load average: 0.00, 0.00, 0.00

Last edited by Wintervenom (2010-08-27 02:49:36)

Bl@ster · 2010-08-27 08:09:22

Wintervenom wrote:

Using Zen 2.6.35 with BFS and BFQ schedulers on an AMD Athlon64 X2 TK-55, an Intel Atom N450, and a VIA Esther C7-D. No load problems so far.

incubine:~> iterate "ssh-cmd ::: uptime" incubine succubine cannabalis
<ssh-rcmd@incubine.incubine.demonpit>  21:48:25 up 20:50,  1 user,  load average: 0.20, 0.31, 0.21
<ssh-rcmd@succubine.incubine.demonpit>  21:48:26 up 59 min,  1 user,  load average: 0.11, 0.34, 0.18
<ssh-rcmd@cannabalis.incubine.demonpit>  21:48:26 up  3:59,  0 users,  load average: 0.00, 0.00, 0.00

It seems to be related only to Intel Dual Core CPUs

dema · 2010-08-27 10:13:23

Bl@ster wrote:

It seems to be related only to Intel Dual Core CPUs

Uhm, I had to downgrade to 2.6.34 too. I had issues with micro-interruption in mouse movements when dragging windows for instance. I operate an eeebox with intel atom N270, intel GMA950.

Cheers

Bl@ster · 2010-08-27 10:18:03

dema wrote:

Bl@ster wrote:
It seems to be related only to Intel Dual Core CPUs
Uhm, I had to downgrade to 2.6.34 too. I had issues with micro-interruption in mouse movements when dragging windows for instance. I operate an eeebox with intel atom N270, intel GMA950.
Cheers

My interruptions are related to windows (especially visibile if wobbly) when I stop the mouse, dragging them. Until I move the cursor, nothin' works, the PC sleeps idle.

karoshi · 2010-08-27 10:30:08

Bl@ster wrote:

It seems to be related only to Intel Dual Core CPUs

hmmm, AMD Athlon XP 2500+ here, also unusual high cpu load

suggy · 2010-08-27 12:12:19

Anyone seen this? https://bugs.launchpad.net/ubuntu/+sour … bug/524281

Mabbe i'm putting 2 and 2 together and getting 5 but could it be related?

Heres my powertop output

Cn                Avg residency       P-states (frequencies)
C0 (cpu running)        (21.4%)         1500 Mhz    25.3%
polling           0.6ms ( 0.0%)         1000 Mhz    74.7%
C1 mwait          0.0ms ( 0.0%)
C2 mwait          0.5ms (14.6%)
C4 mwait          0.9ms (64.0%)

Wakeups-from-idle per second : 974.8    interval: 10.0s
Power usage (5 minute ACPI estimate) :  23.1 W (1.8 hours left)

Top causes for wakeups:
  28.2% (367.9)   PS/2 keyboard/mouse/touchpad interrupt
  18.9% (245.9)   [kernel scheduler] Load balancing tick
  12.4% (161.1)   [extra timer interrupt]
  10.7% (139.5)   firefox
  10.0% (130.6)   [i915@pci:0000:00:02.0] <interrupt>
   8.3% (108.2)   plugin-containe
   2.8% ( 37.1)   [iwl3945] <interrupt>

Running a custom configured kernel, only relevant change from default arch config is that cpu type is changed from 'pentium pro' to 'core 2/newer xeon'.

Last edited by suggy (2010-08-27 13:09:11)

broch · 2010-08-27 13:20:34

HP dv5000 (dual core T2500 nvidia go 7400)
uname -r
2.6.35.1-RIGAUDON+
custom kernel

uptime
06:17:32 up 15 days, 22:31, 2 users, load average: 0.17, 0.19, 0.19

however I think that 2.6.35.x is seriously messed up:
kms/intel video worse than in .34 (I had to go back to 2.6.34 on my netbook )
power management worse than in .34
now this thread regarding I/O issues in .35

Last edited by broch (2010-08-27 13:21:47)

karol · 2010-08-27 13:45:44

broch wrote:

HP dv5000 (dual core T2500 nvidia go 7400)
uname -r
2.6.35.1-RIGAUDON+
custom kernel
uptime
06:17:32 up 15 days, 22:31, 2 users, load average: 0.17, 0.19, 0.19
however I think that 2.6.35.x is seriously messed up:
kms/intel video worse than in .34 (I had to go back to 2.6.34 on my netbook )
power management worse than in .34
now this thread regarding I/O issues in .35

Linux giveth and Linus taketh away ;-)

andrews · 2010-08-28 06:34:30

The problem still present in the testing kernel26 2.6.35.4-1 http://www.archlinux.org/packages/testi … /kernel26/

high load on idle

CD2/Nvidia 9500

Im back to the lts kernel.

litemotiv · 2010-08-28 09:02:32

Here's a bugzilla regression report on it: https://bugzilla.kernel.org/show_bug.cgi?id=16525

It's not very active (yet) though.

dema · 2010-08-29 17:18:49

New package released kernel26 2.6.35.4-1 http://www.archlinux.org/packages/core/i686/kernel26/, problem solved?
Anyone can confirm?

Cheers!

splittercode · 2010-08-29 17:44:57

dema wrote:

New package released kernel26 2.6.35.4-1 http://www.archlinux.org/packages/core/i686/kernel26/, problem solved?
Anyone can confirm?
Cheers!

After ~15 idle minutes I still have a load of 0.48 0.47 0.44. Before I rebooted, while still on LTS, it was normal. So i guess it is not fixed .

note: I use a P4.

Last edited by splittercode (2010-08-29 17:45:37)

archman-cro · 2010-08-29 23:58:25

Hmm, I use i945 and kernel26-zen 2.6.34-1, Celeron M 1.7GHz with 1.2GB RAM.

01:57:49 up 1 day, 21:51,  1 user,  load average: 0.25, 0.33, 0.48

But it varies quickly. Gonna go study what that numbers mean...

karol · 2010-08-30 00:02:01

archman-cro wrote:

Hmm, I use i945 and kernel26-zen 2.6.34-1, Celeron M 1.7GHz with 1.2GB RAM.
01:57:49 up 1 day, 21:51,  1 user,  load average: 0.25, 0.33, 0.48
But it varies quickly. Gonna go study what that numbers mean...

http://en.wikipedia.org/wiki/Load_(computing)

architech · 2010-08-30 00:57:47

Just upgraded to kernel26 2.6.35.4-1 and now the problem is even is worse. On boot and login, the load was 1.42 0.48. 0.30.

Here's the load when idle for 5 minutes.

 00:51:40 up 7 min,  2 users,  load average: 0.42, 0.38, 0.20
Linux 2.6.35-ARCH #1 SMP PREEMPT Fri Aug 27 17:14:28 CEST 2010 x86_64 AMD Athlon(tm) X2 Dual-Core QL-62 AuthenticAMD GNU/Linux

archman-cro · 2010-08-30 08:58:47

Is all of this somehow related to the vanilla-anti-io-stalling patch in kernel26-ice (AUR)? Should I build the kernel with that option?

Arch Linux

#26 2010-08-25 12:31:07

Re: (Solved) Kernel 2.6.35 I/O congestion

#27 2010-08-25 12:48:28

Re: (Solved) Kernel 2.6.35 I/O congestion

#28 2010-08-25 13:20:48

Re: (Solved) Kernel 2.6.35 I/O congestion

#29 2010-08-25 13:22:24

Re: (Solved) Kernel 2.6.35 I/O congestion

#30 2010-08-25 17:30:36

Re: (Solved) Kernel 2.6.35 I/O congestion

#31 2010-08-26 02:58:16

Re: (Solved) Kernel 2.6.35 I/O congestion

#32 2010-08-26 07:19:18

Re: (Solved) Kernel 2.6.35 I/O congestion

#33 2010-08-26 10:44:52

Re: (Solved) Kernel 2.6.35 I/O congestion

#34 2010-08-27 02:07:21

Re: (Solved) Kernel 2.6.35 I/O congestion

#35 2010-08-27 02:48:50

Re: (Solved) Kernel 2.6.35 I/O congestion

#36 2010-08-27 08:09:22

Re: (Solved) Kernel 2.6.35 I/O congestion

#37 2010-08-27 10:13:23

Re: (Solved) Kernel 2.6.35 I/O congestion

#38 2010-08-27 10:18:03

Re: (Solved) Kernel 2.6.35 I/O congestion

#39 2010-08-27 10:30:08

Re: (Solved) Kernel 2.6.35 I/O congestion

#40 2010-08-27 12:12:19

Re: (Solved) Kernel 2.6.35 I/O congestion

#41 2010-08-27 13:20:34

Re: (Solved) Kernel 2.6.35 I/O congestion

#42 2010-08-27 13:45:44

Re: (Solved) Kernel 2.6.35 I/O congestion

#43 2010-08-28 06:34:30

Re: (Solved) Kernel 2.6.35 I/O congestion

#44 2010-08-28 09:02:32

Re: (Solved) Kernel 2.6.35 I/O congestion

#45 2010-08-29 17:18:49

Re: (Solved) Kernel 2.6.35 I/O congestion

#46 2010-08-29 17:44:57

Re: (Solved) Kernel 2.6.35 I/O congestion

#47 2010-08-29 23:58:25

Re: (Solved) Kernel 2.6.35 I/O congestion

#48 2010-08-30 00:02:01

Re: (Solved) Kernel 2.6.35 I/O congestion

#49 2010-08-30 00:57:47

Re: (Solved) Kernel 2.6.35 I/O congestion

#50 2010-08-30 08:58:47

Re: (Solved) Kernel 2.6.35 I/O congestion

Board footer