добрый день уважаемые.
есть фрибс 6.2, на нем (из основного): exim, clam, postgresql, spamassassin (с bayes в постгресе), apache

с недавнего времени стал замечать таймауты при проверке на спам из ексима.
задумался...
посмотрел vmstat, и понял что затык на винте. параметр b до 15-20 доходит.
база bayes ~900Mb (в общем то не много).
средняя скорость поступления сообщений на exim ~20 в минуту (тоже не много)

вот что говорит diskinfo

diskinfo -c /dev/ad0
/dev/ad0
        512             # sectorsize
        61492838400     # mediasize in bytes (57G)
        120103200       # mediasize in sectors
        119150          # Cylinders according to firmware.
        16              # Heads according to firmware.
        63              # Sectors according to firmware.

I/O command overhead:
        time to read 10MB block      4.429599 sec       =    0.216 msec/sector
        time to read 20480 sectors 115.109551 sec       =    5.621 msec/sector
        calculated command overhead                     =    5.404 msec/sector

вот atacontrol

atacontrol mode ad0
current mode = UDMA100

smart вот что говорит

smartctl -a /dev/ad0
smartctl version 5.37 [i386-portbld-freebsd6.2] Copyright (C) 2002-6 Bruce Allen
Home page is http://smartmontools.sourceforge.net/

=== START OF INFORMATION SECTION ===
Model Family:     Maxtor DiamondMax Plus 9 family
Device Model:     Maxtor 6Y060L0
Serial Number:    Y2TGEMQE
Firmware Version: YAR41BW0
User Capacity:    61,492,838,400 bytes
Device is:        In smartctl database [for details use: -P show]
ATA Version is:   7
ATA Standard is:  ATA/ATAPI-7 T13 1532D revision 0
Local Time is:    Sat Aug 18 10:04:45 2007 MSD
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED

General SMART Values:
Offline data collection status:  (0x00) Offline data collection activity
                                        was never started.
                                        Auto Offline Data Collection: Disabled.
Self-test execution status:      ( 243) Self-test routine in progress...
                                        30% of test remaining.
Total time to complete Offline
data collection:                 ( 181) seconds.
Offline data collection
capabilities:                    (0x5b) SMART execute Offline immediate.
                                        Auto Offline data collection on/off support.
                                        Suspend Offline collection upon new
                                        command.
                                        Offline surface scan supported.
                                        Self-test supported.
                                        No Conveyance Self-test supported.
                                        Selective Self-test supported.
SMART capabilities:            (0x0003) Saves SMART data before entering
                                        power-saving mode.
                                        Supports SMART auto save timer.
Error logging capability:        (0x01) Error logging supported.
                                        No General Purpose Logging support.
Short self-test routine
recommended polling time:        (   2) minutes.
Extended self-test routine
recommended polling time:        (  32) minutes.

SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  3 Spin_Up_Time            0x0027   224   224   063    Pre-fail  Always       -       3123
  4 Start_Stop_Count        0x0032   253   253   000    Old_age   Always       -       26
  5 Reallocated_Sector_Ct   0x0033   253   253   063    Pre-fail  Always       -       0
  6 Read_Channel_Margin     0x0001   253   253   100    Pre-fail  Offline      -       0
  7 Seek_Error_Rate         0x000a   253   098   000    Old_age   Always       -       0
  8 Seek_Time_Performance   0x0027   247   242   187    Pre-fail  Always       -       45404
  9 Power_On_Minutes        0x0032   182   182   000    Old_age   Always       -       809h+29m
10 Spin_Retry_Count        0x002b   253   252   157    Pre-fail  Always       -       0
11 Calibration_Retry_Count 0x002b   253   252   223    Pre-fail  Always       -       0
12 Power_Cycle_Count       0x0032   253   253   000    Old_age   Always       -       94
192 Power-Off_Retract_Count 0x0032   253   253   000    Old_age   Always       -       0
193 Load_Cycle_Count        0x0032   253   253   000    Old_age   Always       -       0
194 Temperature_Celsius     0x0032   253   253   000    Old_age   Always       -       48
195 Hardware_ECC_Recovered  0x000a   253   252   000    Old_age   Always       -       5026
196 Reallocated_Event_Count 0x0008   253   253   000    Old_age   Offline      -       0
197 Current_Pending_Sector  0x0008   253   253   000    Old_age   Offline      -       0
198 Offline_Uncorrectable   0x0008   253   253   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x0008   199   199   000    Old_age   Offline      -       0
200 Multi_Zone_Error_Rate   0x000a   253   252   000    Old_age   Always       -       0
201 Soft_Read_Error_Rate    0x000a   253   001   000    Old_age   Always       -       9
202 TA_Increase_Count       0x000a   253   252   000    Old_age   Always       -       0
203 Run_Out_Cancel          0x000b   253   252   180    Pre-fail  Always       -       1
204 Shock_Count_Write_Opern 0x000a   253   252   000    Old_age   Always       -       0
205 Shock_Rate_Write_Opern  0x000a   253   252   000    Old_age   Always       -       0
207 Spin_High_Current       0x002a   253   252   000    Old_age   Always       -       0
208 Spin_Buzz               0x002a   253   252   000    Old_age   Always       -       0
209 Offline_Seek_Performnce 0x0024   193   193   000    Old_age   Offline      -       0
99 Unknown_Attribute       0x0004   253   253   000    Old_age   Offline      -       0
100 Unknown_Attribute       0x0004   253   253   000    Old_age   Offline      -       0
101 Unknown_Attribute       0x0004   253   253   000    Old_age   Offline      -       0

SMART Error Log Version: 1
No Errors Logged

mount
/dev/ad0s1a on / (ufs, local, noatime)
devfs on /dev (devfs, local)
/dev/ad0s1e on /tmp (ufs, local, noatime, soft-updates)
/dev/ad0s1f on /usr (ufs, local, noatime, soft-updates)
/dev/ad0s1d on /var (ufs, local, noatime, soft-updates)
fdescfs on /dev/fd (fdescfs)

df -h
Filesystem     Size    Used   Avail Capacity  Mounted on
/dev/ad0s1a    496M     61M    395M    13%    /
devfs          1.0K    1.0K      0B   100%    /dev
/dev/ad0s1e    496M    7.5M    449M     2%    /tmp
/dev/ad0s1f     53G    3.7G     45G     8%    /usr
/dev/ad0s1d    1.2G    584M    537M    52%    /var
fdescfs        1.0K    1.0K      0B   100%    /dev/fd

ps: async делать не хочу :)

• FreeBSD 6.2: производительность дисковой подсистемы,chaman, 20:02 , 18-Авг-07
  • FreeBSD 6.2: производительность дисковой подсистемы,chocholl, 14:48 , 19-Авг-07

>=== START OF INFORMATION SECTION ===
>Model Family:     Maxtor DiamondMax Plus 9 family

# diskinfo -c /dev/da0
/dev/da0
        512             # sectorsize
        36703934464     # mediasize in bytes (34G)
        71687372        # mediasize in sectors
        4462            # Cylinders according to firmware.
        255             # Heads according to firmware.
        63              # Sectors according to firmware.

I/O command overhead:
        time to read 10MB block      0.204496 sec       =    0.010 msec/sector
        time to read 20480 sectors   2.860629 sec       =    0.140 msec/sector
        calculated command overhead                     =    0.130 msec/sector

# diskinfo -c /dev/da1
/dev/da1
        512             # sectorsize
        36778545152     # mediasize in bytes (34G)
        71833096        # mediasize in sectors
        4471            # Cylinders according to firmware.
        255             # Heads according to firmware.
        63              # Sectors according to firmware.

I/O command overhead:
        time to read 10MB block      0.163460 sec       =    0.008 msec/sector
        time to read 20480 sectors   2.396356 sec       =    0.117 msec/sector
        calculated command overhead                     =    0.109 msec/sector

# diskinfo -c /dev/da2
/dev/da2
        512             # sectorsize
        18351959040     # mediasize in bytes (17G)
        35843670        # mediasize in sectors
        2231            # Cylinders according to firmware.
        255             # Heads according to firmware.
        63              # Sectors according to firmware.

I/O command overhead:
        time to read 10MB block      0.302858 sec       =    0.015 msec/sector
        time to read 20480 sectors   4.331888 sec       =    0.212 msec/sector
        calculated command overhead                     =    0.197 msec/sector

# uname
FreeBSD 6.2

примерно так...
оно робит как веб-(апач+php+perl+tomcat)-мэйл(sendmail+cyrus-imap+clamd+spamassassin)-фтп(proftpd)-сервер

это, видимо, к вопросу IDE vs SCSI via FreeBSD
ничего личного... статистикой поделился просто

>это, видимо, к вопросу IDE vs SCSI via FreeBSD
>ничего личного... статистикой поделился просто

пока возможности переехать нету. я бы и сам рад.

исследуя дисковые операции по процессам, понял, что весь затык на чтении из базы байеса постгресом.
так как чтение происходит намного чаще записи, планирую использовать gmirror для raid1.
поэкспериментирую с параметром balance (load, round-robin или split)

думаю поможет.