大家看看我的checkpoint 为什么这么久阿?

f1

14:19:47  Checkpoint Completed:  duration was 436 seconds.
14:19:47  Checkpoint loguniq 2, logpos 0x5e95018

14:32:26  Checkpoint Completed:  duration was 458 seconds.
14:32:26  Checkpoint loguniq 2, logpos 0x5e96018

14:44:36  Checkpoint Completed:  duration was 430 seconds.
14:44:36  Checkpoint loguniq 2, logpos 0x5e97018

14:56:51  Checkpoint Completed:  duration was 436 seconds.
14:56:51  Checkpoint loguniq 2, logpos 0x5e980b0

15:09:08  Checkpoint Completed:  duration was 437 seconds.
15:09:08  Checkpoint loguniq 2, logpos 0x5e99018

15:21:40  Checkpoint Completed:  duration was 451 seconds.
15:21:40  Checkpoint loguniq 2, logpos 0x5e9a018

15:33:56  Checkpoint Completed:  duration was 437 seconds.
15:33:56  Checkpoint loguniq 2, logpos 0x5e9b018

15:46:28  Checkpoint Completed:  duration was 451 seconds.
15:46:28  Checkpoint loguniq 2, logpos 0x5e9c018

15:58:54  Checkpoint Completed:  duration was 446 seconds.
15:58:54  Checkpoint loguniq 2, logpos 0x5e9d018


onconfig



#**************************************************************************
#
#                           INFORMIX SOFTWARE, INC.
#
#  Title:        onconfig.std
#  Description: Informix Dynamic Server Configuration Parameters
#
#**************************************************************************

# Root Dbspace Configuration

ROOTNAME        rootdbs         # Root dbspace name
ROOTPATH        /home/informix/data/rchunk1
                                # Path for device containing root dbspace
ROOTOFFSET      0               # Offset of root dbspace into device (Kbytes)
ROOTSIZE        2097150         # Size of root dbspace (Kbytes)

# Disk Mirroring Configuration Parameters

MIRROR          0               # Mirroring flag (Yes = 1, No = 0)
MIRRORPATH                      # Path for device containing mirrored root
MIRROROFFSET    0               # Offset into mirrored device (Kbytes)

# Physical Log Configuration

PHYSDBS         rootdbs         # Location (dbspace) of physical log
PHYSFILE        131072          # Physical log file size (Kbytes)

# Logical Log Configuration

LOGFILES        20              # Number of logical log files
LOGSIZE         65536           # Logical log size (Kbytes)

# Diagnostics

MSGPATH         /home/informix/online.log # System message log file path
CONSOLE         /dev/console    # System console message path
ALARMPROGRAM    /home/informix/etc/log_full.sh # Alarm program path
SYSALARMPROGRAM /home/informix/etc/evidence.sh # System Alarm program path
TBLSPACE_STATS  1               

# System Archive Tape Device

TAPEDEV         /dev/rmt/ctape1 # Tape device path       
TAPEBLK         16              # Tape block size (Kbytes)
TAPESIZE        25165800        # Maximum amount of data to put on tape (Kbytes)

# Log Archive Tape Device

LTAPEDEV        /dev/null       # Log tape device path
LTAPEBLK        16              # Log tape block size (Kbytes)
LTAPESIZE       10240           # Max amount of data to put on log tape (Kbytes)

# Optical

STAGEBLOB                       # Informix Dynamic Server/Optical staging area

# System Configuration

SERVERNUM       0               # Unique id corresponding to a Dynamic Server instance
DBSERVERNAME    shpcrbak        # Name of default database server
DBSERVERALIASES                 # List of alternate dbservernames
NETTYPE         ipcshm,1,10,CPU # Configure poll thread(s) for nettype
NETTYPE         tlitcp,5,10,NET # Configure poll thread(s) for nettype
DEADLOCK_TIMEOUT 60              # Max time to wait of lock in distributed env.
RESIDENT        0               # Forced residency flag (Yes = 1, No = 0)

MULTIPROCESSOR  1               # 0 for single-processor, 1 for multi-processor
NUMCPUVPS       2               # Number of user (cpu) vps
SINGLE_CPU_VP   0               # If non-zero, limit number of cpu vps to one

AFF_SPROC       0               # Affinity start processor
AFF_NPROCS      0               # Affinity number of processors

# Shared Memory Parameters

LOCKS           1000000         # Maximum number of locks
BUFFERS         240000          # Maximum number of shared buffers
NUMAIOVPS                       # Number of IO vps
PHYSBUFF        32              # Physical log buffer size (Kbytes)
LOGBUFF         32              # Logical log buffer size (Kbytes)
LOGSMAX         20              # Maximum number of logical log files
CLEANERS        1               # Number of buffer cleaner processes
SHMBASE         0xa000000         # Shared memory base address
SHMVIRTSIZE     8000            # initial virtual shared memory segment size
SHMADD          8192            # Size of new shared memory segments (Kbytes)
SHMTOTAL        0               # Total shared memory (Kbytes). 0=>;unlimited
CKPTINTVL       300             # Check point interval (in sec)
LRUS            8               # Number of LRU queues
LRU_MAX_DIRTY   60              # LRU percent dirty begin cleaning limit
LRU_MIN_DIRTY   50              # LRU percent dirty end cleaning limit
LTXHWM          50              # Long transaction high water mark percentage
LTXEHWM         60              # Long transaction high water mark (exclusive)
TXTIMEOUT       0x12c             # Transaction timeout (in sec)
STACKSIZE       32              # Stack size (Kbytes)

# System Page Size
# BUFFSIZE - Dynamic Server no longer supports this configuration parameter.
#            To determine the page size used by Dynamic Server on your platform
#            see the last line of output from the command, 'onstat -b'.


# Recovery Variables
# OFF_RECVRY_THREADS:
# Number of parallel worker threads during fast recovery or an offline restore.
# ON_RECVRY_THREADS:
# Number of parallel worker threads during an online restore.

OFF_RECVRY_THREADS 10              # Default number of offline worker threads
ON_RECVRY_THREADS 1               # Default number of online worker threads

# Data Replication Variables
# DRAUTO: 0 manual, 1 retain type, 2 reverse type
DRAUTO          0               # DR automatic switchover
DRINTERVAL      30              # DR max time between DR buffer flushes (in sec)
DRTIMEOUT       30              # DR network timeout (in sec)
DRLOSTFOUND     /home/informix/etc/dr.lostfound # DR lost+found file path

# CDR Variables
CDR_LOGBUFFERS  2048            # size of log reading buffer pool (Kbytes)
CDR_EVALTHREADS 1,2             # evaluator threads (per-cpu-vp,additional)
CDR_DSLOCKWAIT  5               # DS lockwait timeout (seconds)
CDR_QUEUEMEM    4096            # Maximum amount of memory for any CDR queue (Kbytes)
CDR_LOGDELTA    30              # % of log space allowed in queue memory
CDR_NUMCONNECT  16              # Expected connections per server
CDR_NIFRETRY    300             # Connection retry (seconds)
CDR_NIFCOMPRESS 0               # Link level compression (-1 never, 0 none, 9 max)

# Backup/Restore variables
BAR_ACT_LOG     /tmp/bar_act.log
BAR_MAX_BACKUP  0               
BAR_RETRY       1               
BAR_NB_XPORT_COUNT 10              
BAR_XFER_BUF_SIZE 31              

# Informix Storage Manager variables
ISM_DATA_POOL   ISMData         # If the data pool name is changed, be sure to
                                # update $INFORMIXDIR/bin/onbar.  Change to
                                # ism_catalog -create_bootstrap -pool <new name>;
ISM_LOG_POOL    ISMLogs         

# Read Ahead Variables
RA_PAGES                        # Number of pages to attempt to read ahead
RA_THRESHOLD                    # Number of pages left before next group

# DBSPACETEMP:
# Dynamic Server equivalent of DBTEMP for SE. This is the list of dbspaces
# that the Dynamic Server SQL Engine will use to create temp tables etc.
# If specified it must be a colon separated list of dbspaces that exist
# when the Dynamic Server system is brought online.  If not specified, or if
# all dbspaces specified are invalid, various ad hoc queries will create
# temporary files in /tmp instead.

DBSPACETEMP     tmpdbs          # Default temp dbspaces

# DUMP*:
# The following parameters control the type of diagnostics information which
# is preserved when an unanticipated error condition (assertion failure) occurs
# during Dynamic Server operations.
# For DUMPSHMEM, DUMPGCORE and DUMPCORE 1 means Yes, 0 means No.

DUMPDIR         /tmp            # Preserve diagnostics in this directory
DUMPSHMEM       1               # Dump a copy of shared memory
DUMPGCORE       0               # Dump a core image using 'gcore'
DUMPCORE        0               # Dump a core image (Warning:this aborts Dynamic Server)
DUMPCNT         1               # Number of shared memory or gcore dumps for
                                # a single user's session

FILLFACTOR      90              # Fill factor for building indexes

# method for Dynamic Server to use when determining current time
USEOSTIME       0               # 0: use internal time(fast), 1: get time from OS(slow)

# Parallel Database Queries (pdq)
MAX_PDQPRIORITY 100             # Maximum allowed pdqpriority
DS_MAX_QUERIES                  # Maximum number of decision support queries
DS_TOTAL_MEMORY                 # Decision support memory (Kbytes)
DS_MAX_SCANS    1048576         # Maximum number of decision support scans       
DATASKIP        off             # List of dbspaces to skip

# OPTCOMPIND
# 0 =>; Nested loop joins will be preferred (where
#      possible) over sortmerge joins and hash joins.
# 1 =>; If the transaction isolation mode is not
#      "repeatable read", optimizer behaves as in (2)
#      below.  Otherwise it behaves as in (0) above.
# 2 =>; Use costs regardless of the transaction isolation
#      mode.  Nested loop joins are not necessarily
#      preferred.  Optimizer bases its decision purely
#      on costs.
OPTCOMPIND      2               # To hint the optimizer

ONDBSPACEDOWN   2               # Dbspace down option: 0 = CONTINUE, 1 = ABORT, 2 = WAIT
LBU_PRESERVE    0               # Preserve last log for log backup
OPCACHEMAX      0               # Maximum optical cache size (Kbytes)

# HETERO_COMMIT (Gateway participation in distributed transactions)
# 1 =>; Heterogeneous Commit is enabled
# 0 (or any other value) =>; Heterogeneous Commit is disabled
HETERO_COMMIT   0               

# Optimization goal: -1 = ALL_ROWS(Default), 0 = FIRST_ROWS
OPT_GOAL        -1              

# Optimizer DIRECTIVES ON (1/Default) or OFF (0)
DIRECTIVES      1               

# Status of restartable restore
RESTARTABLE_RESTORE off

wolfop

你的系统IO那么差么？可以调整调整LRU来减少CHK WRITE，但是你的IO好像也太不好了。

f1

[quote]原帖由 "wolfop"]你的系统IO那么差么？可以调整调整LRU来减少CHK WRITE，但是你的IO好像也太不好了。[/quote 发表：


调整LRU是变大还是变小阿？
IO 差怎么看啊？是不是操作系统配置有问题？我的系统是unixware7.1.3
能不能说详细一点，谢谢！

xjfirst

修改一下你的informix参数，把LRU_MAX_DIRTY设为2，把LRU_MIN_DIRTY设为1；然后将online重启一下，再进行监测。一般情况是你的online设置的有问题！

amiescort

在chkpnt的时候用onstat -F -r 1查看cleaner对chunk的写入情况，如果一直在写入的话，可以44楼上的办法，不过400＋秒的chkpnt，恐怕不太正常。

大梦

LRU_MAX_DIRTY 60 # LRU percent dirty begin cleaning limit
LRU_MIN_DIRTY 50 # LRU percent dirty end cleaning limit
调小些
LRU_MAX_DIRTY 2 # LRU percent dirty begin cleaning limit
LRU_MIN_DIRTY 1 # LRU percent dirty end cleaning limit

看你的情况应该是在跑一些批量处理作业

onstat -R
看看脏页的情况

rock_jq

不正常，PHYSFILE 小点看看

f1

我把以下两个参数修改后
LRU_MAX_DIRTY 60
LRU_MIN_DIRTY 50

同样执行几个大表数据的插入时，checkpoint的时间变短了，最大的也就211 seconds.  

然后我在checkponit 的时候 执行onstat -F -r 1 发现 chunk writes
的数字是一直变大的。

Informix Dynamic Server Version 7.31.UD1    -- On-Line (CKPT REQ) -- Up 01:11:1s
Blocked:CKPT


Fg Writes     LRU Writes    Chunk Writes
0             209126        108548

address  flusher  state    data
2c224510 0        C        12       = 0Xc
      states: Exit Idle Chunk Lru


Informix Dynamic Server Version 7.31.UD1    -- On-Line (CKPT REQ) -- Up 01:11:1s
Blocked:CKPT


Fg Writes     LRU Writes    Chunk Writes
0             209126        108848

address  flusher  state    data
2c224510 0        C        12       = 0Xc
      states: Exit Idle Chunk Lru

大家看看我还能做什么改进？

f1

另外，onstat -R 结果如下：

Informix Dynamic Server Version 7.31.UD1    -- On-Line -- Up 01:30:46 -- 575296s

8 buffer LRU queue pairs                     priority levels
# f/m  pair total  % of    length     LOW   MED_LOW  MED_HIGH   HIGH
0 f     29938    92.0%    27545        0    24624     2870       51
1 m               8.0%     2393        0     2388        5        0
2 f     29955    90.6%    27141        0    24176     2915       50
3 m               9.4%     2814        0     2810        4        0
4 F     29912    94.2%    28187        0    25182     2953       52
5 M               5.8%     1725        0     1723        2        0
6 f     29944    91.4%    27357        0    24516     2785       56
7 m               8.6%     2587        0     2582        5        0
8 f     29949    92.7%    27769        0    24854     2861       54
9 m               7.3%     2180        0     2176        4        0
10 f     29932    89.3%    26715        0    23825     2831       59
11 m              10.7%     3217        0     3208        9        0
12 f     29942    92.1%    27582        0    24689     2841       52
13 m               7.9%     2360        0     2356        4        0
14 f     29926    93.3%    27906        0    25038     2811       57
15 m               6.7%     2020        0     2017        3        0
19296 dirty, 239498 queued, 240000 total, 262144 hash buckets, 2048 buffer size
start clean at 2% (of pair total) dirty, or 600 buffs dirty, stop at 1%
0 priority downgrades, 0 priority upgrades

xjfirst

你还可以试着加大一些BUFFER的大小，同时将LRUS参数调大一些，这些需要根据你自己的硬件配置进行设置。加大BUFFER主要是提高读写命中率，减轻磁盘IO的压力。最好定期整理一下你的数据库，以提高IDS的效率。