Chinaunix
标题:
onstat -p结果的请教
[打印本页]
作者:
StoneIII
时间:
2006-02-21 14:31
标题:
onstat -p结果的请教
以下是onstat -p命令的输出
$ onstat -p
Informix Dynamic Server 2000 Version 9.21.UC3 -- On-Line -- Up 4 days 04:37:24 -- 1553408 Kbytes
Profile
dskreads pagreads bufreads %cached dskwrits pagwrits bufwrits %cached
2322161926 5927888 2251023172 0.00 334399 2267652 8724339 96.17
isamtot open start read write rewrite delete commit rollbk
291578037 12247606 19049368 204321699 2707525 1220681 2857 1642905 8062
gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs
0 0 0 0 0 0 0
ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes
0 0 0 532528.98 22636.56 546 1206
bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans
543003359 2459490 3144619624 12 0 174 126672 338272
ixda-RA idx-RA da-RA RA-pgsused lchwaits
1867330 2 2311008346 2312888202 15352382
看到有些资料说bufreads %cached 正常应超过95%,但该系统在informix刚启动时只能达到70%-80%,然后两三天之内慢慢下降,上星期五到昨天就降到了0%;
另外bufwaits 值也很大。
几乎相同硬件配置和onconfig的另一系统在这两个参数上就表现较好。
目前测试数据库性能还算可以
请问各位朋友这样会不会有什么问题?
作者:
wenlq
时间:
2006-02-21 14:44
很奇怪 你的dskreads 竟然 > bufreads
你的BUFFERS 多大?
作者:
StoneIII
时间:
2006-02-21 14:50
主要配置如下
LOCKS 150000 # Maximum number of locks
BUFFERS 300000 # Maximum number of shared buffers
NUMAIOVPS 2 # Number of IO vps
PHYSBUFF 20240 # Physical log buffer size (Kbytes)
LOGBUFF 20240 # Logical log buffer size (Kbytes)
LOGSMAX 25 # Maximum number of logical log files
CLEANERS 8 # Number of buffer cleaner processes
SHMBASE 0xa000000 # Shared memory base address
SHMVIRTSIZE 800000 # initial virtual shared memory segment size
SHMADD 32768 # Size of new shared memory segments (Kbytes)
SHMTOTAL 2000000 # Total shared memory (Kbytes). 0=>unlimited
CKPTINTVL 600 # 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)
机器物理内存8G
很多相同的业务系统,buffers配置比它小、业务量相同都没有这个情况
我也是觉得很奇怪啊
作者:
jl8323
时间:
2006-02-21 14:54
标题:
回复 1楼 StoneIII 的帖子
UNBUFFER?
把ONCONFIG# Shared Memory Parameters贴出来
作者:
StoneIII
时间:
2006-02-21 15:08
4楼的朋友不知道您指的是哪些参数?
我在上面贴了onconfig的主要配置了
另外onstat -g seg输出如下
Informix Dynamic Server 2000 Version 9.21.UC3 -- On-Line -- Up 4 days 05:22:03 -- 1553408 Kbytes
Segment Summary:
id key addr size ovhd class blkused blkfree
500 1381386241 a000000 770703360 238800 R 187888 272
503 1381386244 37f00000 819986432 25624 V 14066 186126
Total: - - 1590689792 - - 201954 186398
(* segment locked in memory)
另回2楼的朋友,磁盘读写确实很高,onstat -g iof如下
AIO global files:
gfd pathname totalops dskread dskwrite io/s
3 rootdbs 2460 1656 804 0.0
4 log_space 14508 203 14305 0.0
5 phy_space 1776 3 1773 0.0
6 tmp_space 218843 110506 108337 0.6
7 meipbase.001 640472 637148 3324 1.8
8 meipbase.002 1 1 0 0.0
9 meip2.001 2083103 2080236 2867 5.7
10 meip2.002 3466007 3459640 6367 9.5
11 meip2.003 112634575 112632687 1888 308.6
12 meip2.004 183870593 183833943 36650 503.8
13 meip2.005 1 1 0 0.0
14 meip2.006 1 1 0 0.0
15 meip2.007 1 1 0 0.0
作者:
javalee21
时间:
2006-02-21 15:29
呵呵,楼上的是creawor的吧
作者:
jl8323
时间:
2006-02-21 15:31
标题:
回复 5楼 StoneIII 的帖子
onstat -b
onstat -R
从状态来看像是没BUFFER,直接从DISK读
实际物理内存8G,
buffer size是2K吧?现在BUFFER才300000=600M,太小了吧?
扩吧,试试
[
本帖最后由 jl8323 于 2006-2-21 15:33 编辑
]
作者:
StoneIII
时间:
2006-02-21 15:47
回7楼的朋友
$onstat -b
Informix Dynamic Server 2000 Version 9.21.UC3 -- On-Line -- Up 4 days 05:56:02 -- 1553408 Kbytes
Buffers
address userthread flgs pagenum memaddr nslots pgflgs xflgs owner waitlist
b93c1c0 0 84 a43d4f 1696b800 24 1 80 ffffffff 0
bdba4fc 0 84 a43dbb 1cd70000 21 1 80 ffffffff 0
cccf478 0 802 a43f15 31d2c800 21 2001 0 0 0
cd17df8 0 84 a43f26 3237c800 24 1 80 ffffffff 0
cd41cec 0 802 a43f24 32722000 21 1 10 0 0
ce2ed68 0 802 a43f3c 33bbe800 21 2001 10 0 0
3727 modified, 0 resident, 300000 total, 524288 hash buckets, 2048 buffer size
$onstat -R
Informix Dynamic Server 2000 Version 9.21.UC3 -- On-Line -- Up 4 days 05:57:02 -- 1553408 Kbytes
8 buffer LRU queue pairs priority levels
# f/m pair total % of length LOW MED_LOW MED_HIGH HIGH
0 F 37435 98.8% 36968 0 36573 388 7
1 m 1.2% 467 0 467 0 0
2 f 37610 98.8% 37145 0 37035 98 12
3 m 1.2% 465 0 465 0 0
4 f 37518 98.7% 37045 0 36887 144 14
5 m 1.3% 473 0 473 0 0
6 f 37471 98.8% 37033 0 36830 189 14
7 m 1.2% 438 0 438 0 0
8 f 37511 98.6% 37001 0 36820 174 7
9 m 1.4% 510 0 510 0 0
10 f 37445 98.8% 36993 0 36712 268 13
11 m 1.2% 452 0 452 0 0
12 f 37503 98.8% 37058 0 36766 279 13
13 m 1.2% 445 0 445 0 0
14 f 37485 98.7% 37011 0 36810 189 12
15 m 1.3% 474 0 472 2 0
3724 dirty, 299978 queued, 300000 total, 524288 hash buckets, 2048 buffer size
start clean at 60% (of pair total) dirty, or 22500 buffs dirty, stop at 50%
0 priority downgrades, 0 priority upgrades
这两个命令本人没什么研究,不知道上面打印能看出什么问题吗?
buffer size是2K;现在BUFFER确实是300000=600M,
虽然相对物理内存小了点,但是在我们同样的系统中这里的buffers算是最大的了,其他地方的bufreads %cached 都能达到95%以上,这里再低也不该在三天内降到0%,然后几乎都是0,所以很不解。
我会试着增大buffers看看
作者:
jl8323
时间:
2006-02-21 16:27
标题:
回复 8楼 StoneIII 的帖子
对了,楼主,你ONSTAT -Z
再监视一下
作者:
sam_8286
时间:
2006-02-22 09:45
如果说应用感觉不慢的话,觉得是一个bug
欢迎光临 Chinaunix (http://bbs.chinaunix.net/)
Powered by Discuz! X3.2