DIA8305C Memory allocation failure occurred

huyuhui001

Please refer to the following info.
每个 32 位 DB2 实例能够对 4 GB 内存寻址。通常希望将大部分内存给缓冲池专用，但是，AIX、HP-UX 和 Windows 上的内存分段会将最大缓冲池的大小限制在 4 GB 以内。即使是在 32 位世界中拥有十分干净的内存模型的 Solaris 上，用于 DB2 缓冲池的内存也不能超过 3.35 GB；4 GB 内存空间的其余内存必须专用于 DB2 的其它共享内存用途。在 HP-UX 上，32 位 DB2 实例所能够创建的最大缓冲池大约是 800 MB。在 HP-UX 上，只有通过使用 32 位 HP-UX 上的 Memory Windows 来运行多个实例，才能使用 1 GB 以上的缓冲池。在 Windows 上，缓冲池被限制为 3 GB，AIX 上是 1.75 GB，而 Linux 上大约是 1 GB。

In Linux platform, here's a limit of around 800MB for the total size of a database shared memory. You can look at the db cfg parms that make this:buffer pools, lock list,package cache, catalog cache and util_heap and check the sum of all these values. I think you meet this type of limitation. You have no other choice in 32bit DB2. I suggest you increase your buffer pool step by step.

[ 本帖最后由 huyuhui001 于 2007-4-15 13:05 编辑 ]

toms1981

原帖由 huyuhui001 于 2007-4-15 13:02 发表
Please refer to the following info.
每个 32 位 DB2 实例能够对 4 GB 内存寻址。通常希望将大部分内存给缓冲池专用，但是，AIX、HP-UX 和 Windows 上的内存分段会将最大缓冲池的大小限制在 4 GB 以内。即使是 ...

Yes I totally agreed. At the same time, I will do some testing where upgrade one of the instance to 64 bits but not sure will it affect the performance for 32 bits communicate to 64 bits.

Do you have any experience modifying the extended shared memory?

toms1981

Too bad... Extended shared memory doesn't solve the problem. Anyone got others suggestion?

psc2001

failed to reply this article 4times


show the results of db2mtrk -d -v
32bit db2 can use 1.75G memory each intance
show db2 get db cfg
db2 "select * from syscat.bufferpools"
of each database

[ 本帖最后由 psc2001 于 2007-4-16 14:47 编辑 ]

toms1981

yes. I know that 32-bits got limitation on shared memory where only allowed to use up to 1.75 GB. that's why I'm asking whether got alternative solution to solve this issue? I read from other forum where they said can extend the shared memory but it doesn't work when I tried. kindly pls advise.

psc2001

i guess it is impossable to extend the limits of share memory
so what u can do is to reduce the useage of share memory
like  buffpools and so on

or create a new instance
backup & restore to the new instance

this way is how i solve this kind of  problem

psc2001

in my test machine
there is 3 database
each has a 1.2G useage  of memory
at first i put them in db2inst1
i met the problem like u
so i create db2inst2,db2inst3 to move other 2 database in it
now each databse is OK

[ 本帖最后由 psc2001 于 2007-4-16 15:21 编辑 ]

toms1981

its already in 3 different instances for 3 databases.

psc2001

show the results of db2mtrk -d -v

xmltiger

用db2iupdt升级实例到64bit就行啦