Memory Leak

[Paul Howarth]

Hongwei Li wrote:
>>On Thu, Feb 10, 2005 at 10:15:39AM +0100, DafyddHugh wrote:
>> > Hi,
>> >
>> > I am running Fedora Core 3 on a machine with 512Mb memory.
>> >
>> > There are 2 problems
>> >
>> > 1. When the physical memory is exhausted the system tries to use swap,
>>the disk spins and the system becomes unusable. This is under
>>investigation, but any ideas would be appreciated.
>> >
>> > 2. More of an issue. Every few minutes (4 or 5) the available physical
>>memory decreases by 64k, while the cache memory increases by 4k or 8k.
>>This is happening on a very "lean" machine (see simple ps post below).
>>No processes seem to be increasing memory at the same time. Usually the
>>machine also runs httpd, mysqld, popfile in addition to the posted ps.
>>It looks like a memory leak, but I can't find the offending process -
>>any ideas?
>>
>>Which kernel version are you running ?
>>What does the output of free, and slabtop look like ?
>>
>>		Dave
> 
> 
> I have a similar question as 2 above.  My system: 2.6.10-1.741_FC3
> The free command displays the free physical memory decreasing continuesly
> every minute or so:
> 
> # free
>              total       used       free     shared    buffers     cached
> Mem:       1035788    1017952      17836          0     296172      81224
> -/+ buffers/cache:     640556     395232
> Swap:      1052216       8120    1044096
> 
> # free
>              total       used       free     shared    buffers     cached
> Mem:       1035788    1017980      17808          0     296228      81244
> -/+ buffers/cache:     640508     395280
> Swap:      1052216       8120    1044096
> 
> # free
>              total       used       free     shared    buffers     cached
> Mem:       1035788    1018100      17688          0     296260      81300
> -/+ buffers/cache:     640540     395248
> Swap:      1052216       8120    1044096
> 
> Is it normal?  I don't know what will happen if the free memory goes to
> zero, but it seems that, in my system, it never goes to zero.  At some
> point, it jumps up a little.  Please note, my this system is a testing
> system, only 2 regular users were set up, and only 1 user is checking the
> testing emails (plus root's shell).  It was just rebooted 2 days ago:
> 
> # top
> top - 11:51:21 up 1 day, 21:36,  3 users,  load average: 0.00, 0.00, 0.00
> Tasks:  90 total,   1 running,  86 sleeping,   0 stopped,   3 zombie
> Cpu(s):  0.3% us,  0.0% sy,  0.0% ni, 99.3% id,  0.0% wa,  0.3% hi,  0.0% si
> Mem:   1035788k total,  1018564k used,    17224k free,   296420k buffers
> Swap:  1052216k total,     8120k used,  1044096k free,    81360k cached
> 
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> 10847 root      16   0  3632  948  756 R  0.7  0.1   0:00.07 top
>  3438 root      16   0  3940  580  492 S  0.3  0.1   0:17.43 nifd
>  3938 root      16   0  7756 4848 1612 S  0.3  0.5   1:11.28 hald
>     1 root      16   0  2756  560  480 S  0.0  0.1   0:00.82 init
>     2 root      34  19     0    0    0 S  0.0  0.0   0:00.08 ksoftirqd/0
> ...
> 
> Why does it take almost all of 1Gb physical ram?

The kernel will attempt to use any available memory for buffering and 
cacheing, as this makes things run faster. When applications need more 
memory, buffer and cache space is released. So the figure you need to 
look at in the "free" output is the one in the row marked "-/+ 
buffers/cache:", which shows memory usage not in buffers and cache; that 
represents how much memory you're "really" using. In your example above, 
  the last run of "free" actually showed more free memory than the first 
one.

Paul.