Dual-Xeon vs Dual-PIII (Was: Re: vinum in 4.x poor performer?)
Marc G. Fournier
scrappy at hub.org
Tue Feb 8 21:02:53 PST 2005
The more I'm looking at this, the less I can believe my 'issue' is with
vinum ... based on one of my other machines, it just doesn't *feel* right
....
I have two servers that are fairly similar in config ... both running
vinum RAID5 over 4 disks ... one is the Dual-Xeon that I'm finding
"problematic" with 73G Seagate drives, and the other is the Dual-PIII with
36G Seagate drives ...
The reason that I'm finding it hard to believe that my problem is with
vinum is that the Dual-PIII is twice as loaded as the Dual-Xeon, but
hardly seems to break a sweat ...
In fact, out of all my servers (3xDual-PIII, 1xDual-Athlon and
1xDual-Xeon), only the Dual-Xeon doesn't seem to be able to perform ...
Now, out of all of the servers, only the Dual-Xeon, of course, supports
HTT, which I *believe* is disabled, but from dmesg:
Copyright (c) 1992-2004 The FreeBSD Project.
Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
The Regents of the University of California. All rights reserved.
FreeBSD 4.10-STABLE #1: Fri Oct 22 15:06:55 ADT 2004
root at neptune.hub.org:/usr/obj/usr/src/sys/kernel
Timecounter "i8254" frequency 1193182 Hz
CPU: Intel(R) Xeon(TM) CPU 2.40GHz (2392.95-MHz 686-class CPU)
Origin = "GenuineIntel" Id = 0xf27 Stepping = 7
Features=0xbfebfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,DTS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM,PBE>
Hyperthreading: 2 logical CPUs
real memory = 4026466304 (3932096K bytes)
avail memory = 3922362368 (3830432K bytes)
Programming 24 pins in IOAPIC #0
IOAPIC #0 intpin 2 -> irq 0
Programming 24 pins in IOAPIC #1
Programming 24 pins in IOAPIC #2
FreeBSD/SMP: Multiprocessor motherboard: 4 CPUs
cpu0 (BSP): apic id: 0, version: 0x00050014, at 0xfee00000
cpu1 (AP): apic id: 1, version: 0x00050014, at 0xfee00000
cpu2 (AP): apic id: 6, version: 0x00050014, at 0xfee00000
cpu3 (AP): apic id: 7, version: 0x00050014, at 0xfee00000
io0 (APIC): apic id: 8, version: 0x00178020, at 0xfec00000
io1 (APIC): apic id: 9, version: 0x00178020, at 0xfec81000
io2 (APIC): apic id: 10, version: 0x00178020, at 0xfec81400
Preloaded elf kernel "kernel" at 0x80339000.
Warning: Pentium 4 CPU: PSE disabled
Pentium Pro MTRR support enabled
Using $PIR table, 19 entries at 0x800f2f30
Its showing "4 CPUs" ... but:
machdep.hlt_logical_cpus: 1
which, from /usr/src/UPDATING indicates that the HTT "cpus" aren't enabled:
20031022:
Support for HyperThread logical CPUs has now been enabled by
default. As a result, the HTT kernel option no longer exists.
Instead, the logical CPUs are always started so that they can
handle interrupts. However, the extra logical CPUs are prevented
from executing user processes by default. To enable the logical
CPUs, change the value of the machdep.hlt_logical_cpus from 1 to
0. This value can also be set from the loader as a tunable of
the same name.
Finally ... top shows:
last pid: 73871; load averages: 9.76, 9.23, 8.16 up 9+02:02:26 00:57:06
422 processes: 8 running, 413 sleeping, 1 zombie
CPU states: 19.0% user, 0.0% nice, 81.0% system, 0.0% interrupt, 0.0% idle
Mem: 2445M Active, 497M Inact, 595M Wired, 160M Cache, 199M Buf, 75M Free
Swap: 2048M Total, 6388K Used, 2041M Free
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU CPU COMMAND
28298 www 64 0 28136K 12404K CPU2 2 80:59 24.51% 24.51% httpd
69232 excalibur 64 0 80128K 76624K RUN 2 2:55 16.50% 16.50% lisp.run
72879 www 64 0 22664K 9444K RUN 0 0:12 12.94% 12.94% httpd
14154 www 64 0 36992K 22880K RUN 0 55:07 12.70% 12.70% httpd
69758 www 63 0 15400K 8756K RUN 0 0:18 11.87% 11.87% httpd
7553 nobody 2 0 158M 131M poll 0 33:19 8.98% 8.98% nsd
70752 setiathome 2 0 14644K 14084K select 2 0:47 8.98% 8.98% perl
71191 setiathome 2 0 13220K 12804K select 0 0:29 8.40% 8.40% perl
70903 setiathome 2 0 14224K 13676K select 0 0:42 7.37% 7.37% perl
33932 setiathome 2 0 21712K 21144K select 0 2:23 4.59% 4.59% perl
In this case ... 0% idle, 81% in system?
As a comparison the Dual-PIII/vinum server looks like:
last pid: 90614; load averages: 3.64, 2.41, 2.69 up 3+08:45:17 00:59:27
955 processes: 12 running, 942 sleeping, 1 zombie
CPU states: 63.9% user, 0.0% nice, 32.6% system, 3.5% interrupt, 0.0% idle
Mem: 2432M Active, 687M Inact, 563M Wired, 147M Cache, 199M Buf, 5700K Free
Swap: 8192M Total, 12M Used, 8180M Free, 12K In
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU CPU COMMAND
90506 scrappy 56 0 19384K 14428K RUN 0 0:06 22.98% 16.41% postgres
90579 root 57 0 3028K 2156K CPU1 1 0:04 26.23% 14.45% top
90554 pgsql -6 0 12784K 7408K RUN 1 0:04 18.76% 11.87% postgres
90529 pgsql 54 0 14448K 8568K RUN 0 0:03 16.90% 11.28% postgres
90560 scrappy -6 0 97368K 56900K vrlock 1 0:03 18.50% 10.99% postgres
90433 root -6 0 576K 392K piperd 1 0:02 10.47% 7.76% gzip
84754 scrappy 2 0 15508K 8380K sbwait 0 2:41 6.30% 6.30% postgres
90553 root 2 0 98M 99120K select 1 0:01 5.94% 3.76% pg_dump
4621 scrappy 2 0 19544K 11988K sbwait 0 9:36 2.05% 2.05% postgres
The Dual-PIII is running an Oct7th kernel while the Dual-Xeon is Oct22nd
... if that means anything ...
Is there anything I can look at or try? More information I can provide?
Thanks ...
On Wed, 9 Feb 2005, Marc G. Fournier wrote:
> On Wed, 9 Feb 2005, Greg 'groggy' Lehey wrote:
>
>> On Tuesday, 8 February 2005 at 23:21:54 -0400, Marc G. Fournier wrote:
>>>
>>> I have a Dual-Xeon server with 4G of RAM, with its primary file system
>>> consisting of 4x73G SCSI drives running RAID5 using vinum ... the
>>> operating system is currently FreeBSD 4.10-STABLE #1: Fri Oct 22 15:06:55
>>> ADT 2004 ... swap usage is 0% (6149) ... and it performs worse then any of
>>> my other servers, and I have less running on it then the other servers ...
>>>
>>> I also have HTT disabled on this server ... and softupdates enabled on the
>>> file system ...
>>>
>>> That said ... am I hitting limits of software raid or is there something I
>>> should be looking at as far as performance is concerned? Maybe something
>>> I have misconfigured?
>>
>> Based on what you've said, it's impossible to tell. Details would be
>> handy.
>
> Like? I'm not sure what would be useful for this one ... I just sent in my
> current drive config ... something else useful?
>
> systat -v output help:
>
> 4 users Load 4.64 5.58 5.77 Feb 9 00:24
>
> Mem:KB REAL VIRTUAL VN PAGER SWAP PAGER
> Tot Share Tot Share Free in out in out
> Act 1904768 137288 3091620 381128 159276 count
> All 3850780 221996 1078752 605460 pages
> 7921 zfod Interrupts
> Proc:r p d s w Csw Trp Sys Int Sof Flt 242 cow 681 total
> 24 9282 949 8414***** 678 349 8198 566916 wire ahd0
> irq16
> 2527420 act 67 ahd1
> irq17
> 54.6%Sys 0.2%Intr 45.2%User 0.0%Nice 0.0%Idl 608208 inact 157 em0
> irq18
> | | | | | | | | | | 146620 cache 200 clk
> irq0
> ===========================>>>>>>>>>>>>>>>>>>>>> 12656 free 257 rtc
> irq8
> daefr
> Namei Name-cache Dir-cache 7363 prcfr
> Calls hits % hits % react
> 46106 46005 100 13 0 pdwake
> pdpgs
> Disks da0 da1 da2 da3 da4 pass0 pass1 intrn
> KB/t 5.32 9.50 12.52 16.00 9.00 0.00 0.00 204096 buf
> tps 23 2 4 3 1 0 0 1610 dirtybuf
> MB/s 0.12 0.01 0.05 0.04 0.01 0.00 0.00 512000 desiredvnodes
> % busy 3 1 1 1 0 0 0 397436 numvnodes
> 166179 freevnodes
>
> Drives da1 -> da4 are used on the vinum array da0 is just the system drive
> ...
>
> ----
> Marc G. Fournier Hub.Org Networking Services (http://www.hub.org)
> Email: scrappy at hub.org Yahoo!: yscrappy ICQ: 7615664
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to "freebsd-questions-unsubscribe at freebsd.org"
>
----
Marc G. Fournier Hub.Org Networking Services (http://www.hub.org)
Email: scrappy at hub.org Yahoo!: yscrappy ICQ: 7615664
More information about the freebsd-questions
mailing list