On Tue, 2008-11-25 at 11:38 -0500, Stephen Wille Padnos wrote:
> Thomas L Marshall wrote:
>
> >[snip]
> >
> >Thanks again Jeff. The links you provided were just what I needed to
> >compile the missing raid10 and dm- modules that I needed. After I was
> >able to simply copy them to the existing modules directory and depmod
> >did the rest.
> >
> >Any chance these modules could be enabled in the distro?
> >
> >CONFIG_MD_RAID10=m
> >...
> >CONFIG_DM_CRYPT=m
> >CONFIG_DM_SNAPSHOT=m
> >CONFIG_DM_MIRROR=m
> >CONFIG_DM_ZERO=m
> >
> >
> Can you do some comparisons of latency tests with these modules in use
> vs. without? I have identified kjournald (the journalling daemon for
> ext3) as a source of high latencies on some machines, so I'm nervous
> about changing block device related drivers. I'm curious about three cases:
> Non-crypt/non-RAID filesystems, relevant modules not loaded
> Non-crypt/non-RAID filesystems in use, but the relevaln modules loaded
> (maybe on a different filesystem or two, which aren't being exercised)
> Exercising crypt/RAID filesystems during the test.
>
> Thanks
> - Steve
>
Steve,
Thank you for your interest in this discussion. There are so many
variables that a simple comparrison of dm-crypt vs. not is likely not
useful on this machine...
First, please let me say that my configuration is very likely to be not
only somewhat unique (as most are) but also probably very odd to most.
My (somewhat older) system is:
cpu:
Intel(R) Pentium(R) 4 CPU 3.20GHz
meminfo:
MemTotal: 2076768 kB
MemFree: 54760 kB
...
SwapTotal: 8000352 kB
SwapFree: 8000352 kB
md:
Personalities : [linear] [multipath] [raid0] [raid1] [raid6]
[raid5] [raid4] [raid10]
md1 : active raid1 sda4[0] sdb4[1]
199687872 blocks [2/2] [UU]
md0 : active raid1 sda3[0] sdb3[1]
40009792 blocks [2/2] [UU]
The motherboard is an older ASUS P4P800-E, with an Nvidia GeForce 6600
video card and using two 320GB Seagate IDE drives (PATA UDMA/100) in a
raid1 configuration:
/dev/sda1 on /boot type ext3 (rw,noatime)
/dev/mapper/vg0-lv2 on /var type ext3 (rw,noatime)
/dev/mapper/vg0-lv3 on /usr type ext3 (rw,noatime)
/dev/mapper/vg0-lv4 on /opt type ext3 (rw,noatime)
/dev/mapper/vg1-lv5 on /home type ext3 (rw,noatime)
/dev/mapper/vg1-lv6 on /amg type ext3 (rw,noatime)
and:
Disk /dev/sda: 30401 cylinders, 255 heads, 63 sectors/track
Units = cylinders of 8225280 bytes, blocks of 1024 bytes,
counting from 0
Device Boot Start End #cyls #blocks Id System
/dev/sda1 * 0+ 61 62- 497983+ 83 Linux
/dev/sda2 62 559 498 4000185 83 Linux
/dev/sda3 560 5540 4981 40009882+ 83 Linux
/dev/sda4 5541 30400 24860 199687950 83 Linux
(both disks are partitioned identically)
The liberal swap space is somewhat a relic of the installation
which was first installed onto the "swap" partition of the first
disk (hence the 4GB per disk swap), then the raid and logical
volumes were created and debootstrap was used to create the
migrate and install the complete distro, the swap partitions
were shreded and the boot partition was mirrored to the first
partition on the second disk.
Finally, /proc/rtai/scheduler:
RTAI LXRT Real Time Task Scheduler.
Calibrated CPU Frequency: 3207439000 Hz
Calibrated interrupt to scheduler latency: 2688 ns
Calibrated oneshot timer setup_to_firing time: 2010 ns
Number of RT CPUs in system: 1
Real time kthreads in resorvoir (cpu/#): (0/1)
Number of forced hard/soft/hard transitions: traps 0, syscalls 0
Priority Period(ns) FPU Sig State CPU Task HD/SF PID
RT_TASK * TIME
------------------------------------------------------------------------------
1 24305 Yes No 0x5 0:1 1 1 0
dfd50000 0
2 996497 Yes No 0x5 0:1 2 1 0
dff80800 0
TIMED
> dfd50000 > dff80800
READY
additional notes:
* Rsync is used for incremental backups to a central server
twice daily and would likely negatively impact latency.
* The system is intended for remote access only via ssh. (i.e.
ssh -Y <remote-machine> latency-test)
* The display manager is NOT running on this system, nor does
this system have an attached monitor or keyboard/mouse.
* Luks keys are stored on a flash drive (for remote admin)
without which the user/company data on the drives is not
accessible.
Now that we have had a look under the hood, the latency numbers will
hopefully become more meaningful (or at least cast in perspective).
Max Interval(ns) Max Jitter(ns) Last interval(ns)
Servo thread (1.0ms): 1003227 6759 996415
Base thread (25.0us): 34090 10024 24310
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
