 |
|
 |
|
On Fri, 27 Jan 2012 16:52:20 +0000, Andy Whitcroft <...@canonical.com
I would expect to find that in older kernels other threads are taking on
the load. Presumably they are flushing pages like mad to make space in
cache for your workload. I guess I am asking, is the overall load
actually higher, or just moved.
-apw
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
 |
|
 |
|
|
|
|
On Mon, 30 Jan 2012 14:06:45 +0100, Christian Brandes <...@forschungsgruppe.de
Hi Andy,
thanks for your first try to explain.
But on 3.0.0-13-generic my "dd" command makes about 75 MByte/s all the time.
On 3.2.0-10-generic it is 98 MByte/s until the cache is filled up and kworker captures my CPU. Then it continues at about 5MByte/s.
It is definitely not just that the load has moved. It is actually multiples higer.
Could you just try to reproduce that on a precise system?
Thanks!
Best regards
Christian
--- original mail: ---
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Thu, 02 Feb 2012 14:35:07 +0100, Christian Brandes <...@forschungsgruppe.de
Hi Andy,
could you reproduce the phenomenon?
Best regards
Christian
--- original mail: ---
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Mon, 06 Feb 2012 10:53:04 +0100, Christian Brandes <...@forschungsgruppe.de
Hi all!
This takes just 5 minutes!!!!!
Please, could someone help me!
It looks like a serious bug in 3.2-Kernels.
Could someone just do "dd if=/dev/zero of=/tmp/zero1.tmp bs=1024 count=13000000" on a precise system with 3.2. kernel and watch CPU usage of kworker threads going up when cache memory is filled up.
Thanks and best regards
Christian
--- original mail: ---
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Mon, 6 Feb 2012 11:06:48 +0000, Andy Whitcroft <...@canonical.com
Nope could not reproduce it here on a machine with only 2GB of ram; this
is with a 3.2.0-12 Ubuntu kernel. I see the expected kswapd0 and flush-8:0
using some 10% between them worst case. Which kernel are you using,
what platform is triggering it.
-apw
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Mon, 06 Feb 2012 14:33:07 +0100, Stefan Bader <...@canonical.com
Same here with 3.2.0-14.23. About 80M/s and same kswapd0 and flush-8:0
percentages. Using a dual core cpu.
One thing to verify as well: is /tmp part of / or a tmpfs mount? (though in that
case one would not expect a 13G write to succeed)...
-Stefan
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Mon, 6 Feb 2012 14:24:30 +0000, Andy Whitcroft <...@canonical.com
Indeed. Also what filesystem type are you using. My testing was with
ext4.
-apw
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|
|
|
On Tue, 07 Feb 2012 12:53:17 +0100, Christian Brandes <...@forschungsgruppe.de
Hi Andy and Stefan,
I am glad! Thanks for your help!
Yes, /tmp is on / and
no, / is not ext4, but xfs.
I installed Ubuntu 12.04 on two physical machines:
HP Compaq 6005 Pro Business PCs
But I tested it on oder Hardware, and had the same result, as well.
8,6 MB/s is very poor, even for a desktop hardware!
That is only the average. The first 8GB go in a few seconds at 95 MB/s and the rest only at a bit more then 1 MB/s.
What can the problem be?
Can it be that xfs on 3.2 gets so badly slow when it fills up more than 75%?
Now I tested with 10GB to not fill the filesystem too much, as I know very full filesystems get slow. But they should not at 80%. And xfs does not on 3.0 kernels.
Best regrads
Christian
----------------------------------------
~# /proc/version
Linux version 3.2.0-12-generic (buildd@crested) (gcc version 4.6.2 (Ubuntu/Linaro 4.6.2-12ubuntu1) ) #21-Ubuntu SMP Tue Jan 31 18:48:57 UTC 2012
The first 8 GB go in a few seconds and make about 50-80 MB/s. Top and bwm-ng look like that:
~# top
top - 12:01:06 up 1 min, 2 users, load average: 1.64, 0.68, 0.25
Tasks: 121 total, 1 running, 120 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.3%us, 6.9%sy, 0.0%ni, 0.7%id, 91.9%wa, 0.0%hi, 0.2%si, 0.0%st
Mem: 1791108k total, 1237680k used, 553428k free, 1144k buffers
Swap: 7812092k total, 0k used, 7812092k free, 1101664k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2067 root 20 0 11500 632 516 D 15 0.0 0:01.64 dd
1352 root 20 0 0 0 0 D 1 0.0 0:00.08 flush-8:0
11 root 20 0 0 0 0 S 1 0.0 0:00.06 kworker/0:1
1 root 20 0 24160 2192 1264 S 0 0.1 0:00.46 init
2 root 20 0 0 0 0 S 0 0.0 0:00.00 kthreadd
3 root 20 0 0 0 0 S 0 0.0 0:00.00 ksoftirqd/0
4 root 20 0 0 0 0 S 0 0.0 0:00.00 kworker/0:0
5 root 20 0 0 0 0 S 0 0.0 0:00.20 kworker/u:0
~# bwm-ng
bwm-ng v0.6 (probing every 0.500s), press 'h' for help
input: libstatdisk type: avg (30s)
| iface Rx Tx Total
==============================================================================
sda: 26.61 KB/s 95667.68 KB/s 95694.29 KB/s
------------------------------------------------------------------------------
total: 26.61 KB/s 95667.68 KB/s 95694.29 KB/s
Then only 1-2 MB/s and top and bwm-ng look like that:
~# top
top - 12:02:26 up 2 min, 2 users, load average: 1.75, 0.96, 0.38
Tasks: 119 total, 2 running, 117 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.8%us, 11.2%sy, 0.0%ni, 53.6%id, 34.2%wa, 0.0%hi, 0.2%si, 0.0%st
Mem: 1791108k total, 1720964k used, 70144k free, 716k buffers
Swap: 7812092k total, 4k used, 7812088k free, 1556516k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4 root 20 0 0 0 0 S 47 0.0 0:04.04 kworker/0:0
9 root 20 0 0 0 0 R 45 0.0 0:04.07 kworker/1:0
2067 root 20 0 11500 632 516 D 4 0.0 0:24.56 dd
2098 root 20 0 17216 1248 908 R 2 0.1 0:00.01 top
1 root 20 0 24160 2080 1152 S 0 0.1 0:00.46 init
2 root 20 0 0 0 0 S 0 0.0 0:00.00 kthreadd
3 root 20 0 0 0 0 S 0 0.0 0:00.00 ksoftirqd/0
5 root 20 0 0 0 0 S 0 0.0 0:00.20 kworker/u:0
~# bwm-ng
bwm-ng v0.6 (probing every 0.500s), press 'h' for help
input: libstatdisk type: avg (30s)
\ iface Rx Tx Total
==============================================================================
sda: 0.00 KB/s 1471.86 KB/s 1471.86 KB/s
------------------------------------------------------------------------------
total: 0.00 KB/s 1471.86 KB/s 1471.86 KB/s
~# dd if=/dev/zero of=/tmp/zero1.tmp bs=1024 count=10000000
10000000+0 records in
10000000+0 records out
10240000000 bytes (10 GB) copied, 1196,41 s, 8,6 MB/s
~# df -m /tmp
Filesystem 1M-blocks Used Available Use% Mounted on
/dev/sda5 15248 12312 2937 81% /
~# mount
/dev/sda5 on / type xfs (rw)
<----------
With 3.0.0 all 10GB go in a few seconds.
~# cat /proc/version
Linux version 3.0.0-13-generic (buildd@crested) (gcc version 4.6.1 (Ubuntu/Linaro 4.6.1-9ubuntu3) ) #22-Ubuntu SMP Wed Nov 2 13:27:26 UTC 2011
~# top
top - 11:51:19 up 1 min, 2 users, load average: 0.62, 0.21, 0.08
Tasks: 113 total, 3 running, 110 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.8%us, 13.6%sy, 0.0%ni, 5.1%id, 80.4%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 1794104k total, 1730888k used, 63216k free, 504k buffers
Swap: 7812092k total, 0k used, 7812092k free, 1576080k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2134 root 20 0 11500 632 516 R 27 0.0 0:04.10 dd
1341 root 20 0 0 0 0 D 3 0.0 0:00.48 flush-8:0
10 root 20 0 0 0 0 S 2 0.0 0:00.24 kworker/0:1
46 root 20 0 0 0 0 R 1 0.0 0:00.04 kswapd0
2136 root 20 0 0 0 0 S 0 0.0 0:00.06 kworker/0:2
~# bwm-ng
bwm-ng v0.6 (probing every 0.500s), press 'h' for help
input: libstatdisk type: avg (30s)
| iface Rx Tx Total
==============================================================================
sda: 0.00 KB/s 91107.56 KB/s 91107.56 KB/s
------------------------------------------------------------------------------
total: 0.00 KB/s 91107.56 KB/s 91107.56 KB/s
~# dd if=/dev/zero of=/tmp/zero1.tmp bs=1024 count=10000000
10000000+0 records in
10000000+0 records out
10240000000 bytes (10 GB) copied, 107,759 s, 95,0 MB/s
/proc/cpuinfo:
--------------
processor : 0
vendor_id : AuthenticAMD
cpu family : 16
model : 6
model name : AMD Athlon(tm) II X2 B24 Processor
stepping : 3
microcode : 0x10000b6
cpu MHz : 800.000
cache size : 1024 KB
physical id : 0
siblings : 2
core id : 0
cpu cores : 2
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips : 5985.14
TLB size : 1024 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate
processor : 1
vendor_id : AuthenticAMD
cpu family : 16
model : 6
model name : AMD Athlon(tm) II X2 B24 Processor
stepping : 3
microcode : 0x10000b6
cpu MHz : 800.000
cache size : 1024 KB
physical id : 0
siblings : 2
core id : 1
cpu cores : 2
apicid : 1
initial apicid : 1
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips : 5985.02
TLB size : 1024 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate
--
kernel-team mailing list
kern...@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/kernel-team
|
|
|
|
|
|
|