关于DB2代理进程占用大量内存资源的问题
问题描述：
AIX5.3 DB2V9.1 HACMP HADR
首先情况topas的信息
说明：分页空间使用率运行三年来从未超过15%，突然从8月20日开始持续上涨
再看ps aux的输出
USER PID %CPU %MEM SZ RSS TTY STAT STIME TIME COMMAND
mesdbadm 336090 0.0 31.0 4205120 2533596 - A Aug 14 19:50 db2agent (DBMES
db2的代理进程占用大量资源
然后是针对PID 336090 的svmon -G的命令输出
Pid Command Inuse Pin Pgsp Virtual 64-bit Mthrd 16MB
336090 db2sysc 1659362 8385 425400 2084244 Y N N
PageSize Inuse Pin Pgsp Virtual
s 4 KB 643010 8337 425368 1067860
m 64 KB 63522 3 2 63524
Vsid Esid Type Description PSize Inuse Pin Pgsp Virtual
30684 7800002a work default shmat/mmap m 4096 0 0 4096
e867f 78000025 work default shmat/mmap m 4096 0 0 4096
28687 7800002d work default shmat/mmap m 4096 0 0 4096
e067e 78000024 work default shmat/mmap m 4096 0 0 4096
50688 7800002e work default shmat/mmap m 4096 0 0 4096
f867d 78000023 work default shmat/mmap m 4096 0 0 4096
8683 78000029 work default shmat/mmap m 4096 0 0 4096
20686 7800002c work default shmat/mmap m 4096 0 0 4096
f067c 78000022 work default shmat/mmap m 4096 0 0 4096
e2559 20 work text data BSS heap s 65536 0 0 65536
682 78000028 work default shmat/mmap m 4096 0 0 4096
38685 7800002b work default shmat/mmap m 4096 0 0 4096
20de6 1f work text data BSS heap s 65536 0 0 65536
18681 78000027 work default shmat/mmap m 4096 0 0 4096
10680 78000026 work default shmat/mmap m 4096 0 0 4096
6a528 18 work text data BSS heap s 65535 0 3 65536
1a186 1b work text data BSS heap s 65535 0 1 65536
12567 1d work text data BSS heap s 65534 0 2 65536
1a546 1c work text data BSS heap s 65534 0 2 65536
d24df 1e work text data BSS heap s 65525 0 11 65536
788ed 1a work text data BSS heap s 65507 0 29 65536
4a56c 19 work text data BSS heap s 65481 0 55 65536
58689 7800002f work default shmat/mmap m 2949 0 0 2949
88733 78000042 work default shmat/mmap m 2333 0 1 2334
e8dff 17 work text data BSS heap s 36689 0 28848 65536
90690 78000036 work default shmat/mmap m 1897 0 0 1897
870b0 90000000 work shared library text m 1442 0 1 1443
b85b5 78000002 work default shmat/mmap m 923 0 0 923
6868f 78000035 work default shmat/mmap m 649 0 0 649
0 0 work kernel segment s 8581 8337 3883 12672
4a42c 21 work text data BSS heap s 4229 0 0 4229
4856b 11 work text data BSS heap s 2097 0 61588 63536
parent=e875f
e857f 78000000 work default shmat/mmap m 58 0 0 58
6056e 77000000 work default shmat/mmap s 900 0 637 1537
e821f 90020014 work shared library s 193 0 525 629
d8759 f00000002 work process private m 12 3 0 12
parent=7874d
8001 9ffffffd work shared library s 148 0 2000 2070
28587 78000001 work default shmat/mmap m 7 0 0 7
608ee 15 work text data BSS heap s 112 0 65520 65536
9a516 14 work text data BSS heap s 101 0 65536 65536
82515 16 work text data BSS heap s 82 0 65535 65536
20766 ffffffff work application stack m 4 0 0 4
parent=10760
c875b 80020014 work USLA heap s 48 0 45 48
parent=762
c075a 9001000a work shared library data s 46 0 47 56
parent=f875d
30026 9ffffffe work shared library s 33 0 20 34
c021a 9fffffff clnt USLA text,/dev/hd2:282 s 11 0 - -
30764 8fffffff work private load data s 8 0 9 9
parent=f075c
8583 10 clnt text data BSS heap, s 8 0 - -
/dev/mesdbadm:113
4806b - clnt /dev/hd2:24006 s 1 0 - -
385c5 78000012 work default shmat/mmap m 0 0 0 0
b8695 7800003b work default shmat/mmap m 0 0 0 0
c869b 78000041 work default shmat/mmap m 0 0 0 0
c21dd 13 work text data BSS heap s 0 0 65536 65536
85c3 78000010 work default shmat/mmap m 0 0 0 0
c05ba 78000007 work default shmat/mmap m 0 0 0 0
585c9 78000016 work default shmat/mmap m 0 0 0 0
485cb 78000018 work default shmat/mmap m 0 0 0 0
985d1 7800001e work default shmat/mmap m 0 0 0 0
4068a 78000030 work default shmat/mmap m 0 0 0 0
885d3 78000020 work default shmat/mmap m 0 0 0 0
d85b9 78000006 work default shmat/mmap m 0 0 0 0
7868d 78000033 work default shmat/mmap m 0 0 0 0
205c6 78000013 work default shmat/mmap m 0 0 0 0
d0698 7800003e work default shmat/mmap m 0 0 0 0
80692 78000038 work default shmat/mmap m 0 0 0 0
805d2 7800001f work default shmat/mmap m 0 0 0 0
405ca 78000017 work default shmat/mmap m 0 0 0 0
605ce 7800001b work default shmat/mmap m 0 0 0 0
c85bb 78000008 work default shmat/mmap m 0 0 0 0
a05b6 78000003 work default shmat/mmap m 0 0 0 0
105c0 7800000d work default shmat/mmap m 0 0 0 0
d24ff 12 work text data BSS heap s 0 0 65536 65536
c069a 78000040 work default shmat/mmap m 0 0 0 0
98691 78000037 work default shmat/mmap m 0 0 0 0
f85bd 7800000a work default shmat/mmap m 0 0 0 0
e05be 7800000b work default shmat/mmap m 0 0 0 0
285c7 78000014 work default shmat/mmap m 0 0 0 0
d8699 7800003f work default shmat/mmap m 0 0 0 0
505c8 78000015 work default shmat/mmap m 0 0 0 0
685cf 7800001c work default shmat/mmap m 0 0 0 0
6068e 78000034 work default shmat/mmap m 0 0 0 0
b0694 7800003a work default shmat/mmap m 0 0 0 0
a85b7 78000004 work default shmat/mmap m 0 0 0 0
4868b 78000031 work default shmat/mmap m 0 0 0 0
a0696 7800003c work default shmat/mmap m 0 0 0 0
e85bf 7800000c work default shmat/mmap m 0 0 0 0
5c2 7800000f work default shmat/mmap m 0 0 0 0
305c4 78000011 work default shmat/mmap m 0 0 0 0
905d0 7800001d work default shmat/mmap m 0 0 0 0
88693 78000039 work default shmat/mmap m 0 0 0 0
705cc 78000019 work default shmat/mmap m 0 0 0 0
d05b8 78000005 work default shmat/mmap m 0 0 0 0
7068c 78000032 work default shmat/mmap m 0 0 0 0
b05d4 78000021 work default shmat/mmap m 0 0 0 0
a8697 7800003d work default shmat/mmap m 0 0 0 0
f05bc 78000009 work default shmat/mmap m 0 0 0 0
785cd 7800001a work default shmat/mmap m 0 0 0 0
185c1 7800000e work default shmat/mmap m 0 0 0 0
请教一下分页空间怎么就突然上去了，而且在持续上涨？内存泄漏？还是别的什么原因？
AIX5.3 DB2V9.1 HACMP HADR
首先情况topas的信息
说明：分页空间使用率运行三年来从未超过15%，突然从8月20日开始持续上涨
再看ps aux的输出
USER PID %CPU %MEM SZ RSS TTY STAT STIME TIME COMMAND
mesdbadm 336090 0.0 31.0 4205120 2533596 - A Aug 14 19:50 db2agent (DBMES
db2的代理进程占用大量资源
然后是针对PID 336090 的svmon -G的命令输出
Pid Command Inuse Pin Pgsp Virtual 64-bit Mthrd 16MB
336090 db2sysc 1659362 8385 425400 2084244 Y N N
PageSize Inuse Pin Pgsp Virtual
s 4 KB 643010 8337 425368 1067860
m 64 KB 63522 3 2 63524
Vsid Esid Type Description PSize Inuse Pin Pgsp Virtual
30684 7800002a work default shmat/mmap m 4096 0 0 4096
e867f 78000025 work default shmat/mmap m 4096 0 0 4096
28687 7800002d work default shmat/mmap m 4096 0 0 4096
e067e 78000024 work default shmat/mmap m 4096 0 0 4096
50688 7800002e work default shmat/mmap m 4096 0 0 4096
f867d 78000023 work default shmat/mmap m 4096 0 0 4096
8683 78000029 work default shmat/mmap m 4096 0 0 4096
20686 7800002c work default shmat/mmap m 4096 0 0 4096
f067c 78000022 work default shmat/mmap m 4096 0 0 4096
e2559 20 work text data BSS heap s 65536 0 0 65536
682 78000028 work default shmat/mmap m 4096 0 0 4096
38685 7800002b work default shmat/mmap m 4096 0 0 4096
20de6 1f work text data BSS heap s 65536 0 0 65536
18681 78000027 work default shmat/mmap m 4096 0 0 4096
10680 78000026 work default shmat/mmap m 4096 0 0 4096
6a528 18 work text data BSS heap s 65535 0 3 65536
1a186 1b work text data BSS heap s 65535 0 1 65536
12567 1d work text data BSS heap s 65534 0 2 65536
1a546 1c work text data BSS heap s 65534 0 2 65536
d24df 1e work text data BSS heap s 65525 0 11 65536
788ed 1a work text data BSS heap s 65507 0 29 65536
4a56c 19 work text data BSS heap s 65481 0 55 65536
58689 7800002f work default shmat/mmap m 2949 0 0 2949
88733 78000042 work default shmat/mmap m 2333 0 1 2334
e8dff 17 work text data BSS heap s 36689 0 28848 65536
90690 78000036 work default shmat/mmap m 1897 0 0 1897
870b0 90000000 work shared library text m 1442 0 1 1443
b85b5 78000002 work default shmat/mmap m 923 0 0 923
6868f 78000035 work default shmat/mmap m 649 0 0 649
0 0 work kernel segment s 8581 8337 3883 12672
4a42c 21 work text data BSS heap s 4229 0 0 4229
4856b 11 work text data BSS heap s 2097 0 61588 63536
parent=e875f
e857f 78000000 work default shmat/mmap m 58 0 0 58
6056e 77000000 work default shmat/mmap s 900 0 637 1537
e821f 90020014 work shared library s 193 0 525 629
d8759 f00000002 work process private m 12 3 0 12
parent=7874d
8001 9ffffffd work shared library s 148 0 2000 2070
28587 78000001 work default shmat/mmap m 7 0 0 7
608ee 15 work text data BSS heap s 112 0 65520 65536
9a516 14 work text data BSS heap s 101 0 65536 65536
82515 16 work text data BSS heap s 82 0 65535 65536
20766 ffffffff work application stack m 4 0 0 4
parent=10760
c875b 80020014 work USLA heap s 48 0 45 48
parent=762
c075a 9001000a work shared library data s 46 0 47 56
parent=f875d
30026 9ffffffe work shared library s 33 0 20 34
c021a 9fffffff clnt USLA text,/dev/hd2:282 s 11 0 - -
30764 8fffffff work private load data s 8 0 9 9
parent=f075c
8583 10 clnt text data BSS heap, s 8 0 - -
/dev/mesdbadm:113
4806b - clnt /dev/hd2:24006 s 1 0 - -
385c5 78000012 work default shmat/mmap m 0 0 0 0
b8695 7800003b work default shmat/mmap m 0 0 0 0
c869b 78000041 work default shmat/mmap m 0 0 0 0
c21dd 13 work text data BSS heap s 0 0 65536 65536
85c3 78000010 work default shmat/mmap m 0 0 0 0
c05ba 78000007 work default shmat/mmap m 0 0 0 0
585c9 78000016 work default shmat/mmap m 0 0 0 0
485cb 78000018 work default shmat/mmap m 0 0 0 0
985d1 7800001e work default shmat/mmap m 0 0 0 0
4068a 78000030 work default shmat/mmap m 0 0 0 0
885d3 78000020 work default shmat/mmap m 0 0 0 0
d85b9 78000006 work default shmat/mmap m 0 0 0 0
7868d 78000033 work default shmat/mmap m 0 0 0 0
205c6 78000013 work default shmat/mmap m 0 0 0 0
d0698 7800003e work default shmat/mmap m 0 0 0 0
80692 78000038 work default shmat/mmap m 0 0 0 0
805d2 7800001f work default shmat/mmap m 0 0 0 0
405ca 78000017 work default shmat/mmap m 0 0 0 0
605ce 7800001b work default shmat/mmap m 0 0 0 0
c85bb 78000008 work default shmat/mmap m 0 0 0 0
a05b6 78000003 work default shmat/mmap m 0 0 0 0
105c0 7800000d work default shmat/mmap m 0 0 0 0
d24ff 12 work text data BSS heap s 0 0 65536 65536
c069a 78000040 work default shmat/mmap m 0 0 0 0
98691 78000037 work default shmat/mmap m 0 0 0 0
f85bd 7800000a work default shmat/mmap m 0 0 0 0
e05be 7800000b work default shmat/mmap m 0 0 0 0
285c7 78000014 work default shmat/mmap m 0 0 0 0
d8699 7800003f work default shmat/mmap m 0 0 0 0
505c8 78000015 work default shmat/mmap m 0 0 0 0
685cf 7800001c work default shmat/mmap m 0 0 0 0
6068e 78000034 work default shmat/mmap m 0 0 0 0
b0694 7800003a work default shmat/mmap m 0 0 0 0
a85b7 78000004 work default shmat/mmap m 0 0 0 0
4868b 78000031 work default shmat/mmap m 0 0 0 0
a0696 7800003c work default shmat/mmap m 0 0 0 0
e85bf 7800000c work default shmat/mmap m 0 0 0 0
5c2 7800000f work default shmat/mmap m 0 0 0 0
305c4 78000011 work default shmat/mmap m 0 0 0 0
905d0 7800001d work default shmat/mmap m 0 0 0 0
88693 78000039 work default shmat/mmap m 0 0 0 0
705cc 78000019 work default shmat/mmap m 0 0 0 0
d05b8 78000005 work default shmat/mmap m 0 0 0 0
7068c 78000032 work default shmat/mmap m 0 0 0 0
b05d4 78000021 work default shmat/mmap m 0 0 0 0
a8697 7800003d work default shmat/mmap m 0 0 0 0
f05bc 78000009 work default shmat/mmap m 0 0 0 0
785cd 7800001a work default shmat/mmap m 0 0 0 0
185c1 7800000e work default shmat/mmap m 0 0 0 0
请教一下分页空间怎么就突然上去了，而且在持续上涨？内存泄漏？还是别的什么原因？
7同行回答
我再问最后一个问题，依我对自己维护的程序的了解 都是一些前后台接受消息的处理， 应该不可能有什么样的程序会使得代理进程占用这么多的内存（63522*64K=3.8G, 643010*4K=2.4G）啊 出现这么大的内存占用量这本身我是没想明白（换页我明白了）
补充一点：
我用svmon -G查看 物理内存几乎是用光了的收起
补充一点：
我用svmon -G查看 物理内存几乎是用光了的收起
浏览883
63522*64K=3.8G, 643010*4K=2.4G
so 3.8+2.4=6.2G
you totally have 7.9G RAM, so that's exactly what i mentioned, system start paging even if there's still free RAM~~~收起
so 3.8+2.4=6.2G
you totally have 7.9G RAM, so that's exactly what i mentioned, system start paging even if there's still free RAM~~~收起
浏览835
我再问一下，用64K就用呗 为什么会占用那么多呢？而且vmstat -P的输出64K占用也不多啊
PageSize Inuse Pin Pgsp Virtual
s 4 KB 643010 8337 425368 1067860
m 64 KB 63522 3 2 63524
还有版主是不是英语水平比汉语还好啊
想了一下，版主的意思是不是就是说 当DB2用到了64K的时候，就可能会有如我所述的那种现象出现？而这种现象就是AIX的一个BUG(还是数据库的BUG？) 这样理解对么？收起
PageSize Inuse Pin Pgsp Virtual
s 4 KB 643010 8337 425368 1067860
m 64 KB 63522 3 2 63524
还有版主是不是英语水平比汉语还好啊
想了一下，版主的意思是不是就是说 当DB2用到了64K的时候，就可能会有如我所述的那种现象出现？而这种现象就是AIX的一个BUG(还是数据库的BUG？) 这样理解对么？收起
浏览803
it's not from db2 level, but from OS level.
in AIX it support 3 different page size, 4K, 64K and 16MB (or something like that, can't remember). By default db2 uses small/medium page size (4K and 64K). There are some APARs in AIX may cause paging when applications use 64K page even when there's still sufficient free RAM. Just disable 64K page and see whether if it helps收起
in AIX it support 3 different page size, 4K, 64K and 16MB (or something like that, can't remember). By default db2 uses small/medium page size (4K and 64K). There are some APARs in AIX may cause paging when applications use 64K page even when there's still sufficient free RAM. Just disable 64K page and see whether if it helps收起
浏览808
浏览778
try to disable medium page support first, sometime it resolves issue like charm, unless you have touched memory configuration or hardware... :)
(note you need to disable both shared/private medium page, so you need 2 steps, see doc for details)
https://www-304.ibm.com/support/docview.wss?uid=swg21417752收起
(note you need to disable both shared/private medium page, so you need 2 steps, see doc for details)
https://www-304.ibm.com/support/docview.wss?uid=swg21417752收起
浏览869
well then maybe you wanna do some memory tuning for the database, for example cap the INSTANCE_MEMORY/DATABASE_MEMORY for the total amount of memory used by the instance/database...
you can use db2pd -memset/ db2pd -mempool / db2pd -db -memset / db2pd -db -mempool
to get memory usage for each indiviual memory pool and see where can be improved...收起
you can use db2pd -memset/ db2pd -mempool / db2pd -db
to get memory usage for each indiviual memory pool and see where can be improved...收起
浏览799