Allocate lots of memory => core dump

Wesley T. Perdue wes at
Mon Jan 20 16:48:01 CST 2003


We have had a number of individual processes crash when they try to allocate more than 3 GB of memory.  We're using dual Xeon systems with 4 GB RAM running Red Hat 7.3 and kernel version 2.4.9-21.1smp.

We knew our memory allocations would grow further, so we moved that application to a Sun Fire v480 with 16 GB RAM, since the app is only available on x86 Linux and Solaris.


At 03:18 PM 1/20/2003 -0700, Eric Swenson wrote:
>I'm sure that there are people here more knowledgable with I, but I believe 
>that the per-process memory limit is 3GB with 2.4 kernels, Windows XP pro, or 
>win2k adv. server, and only 2GB per process with 2.2 kernels, Windows 2k Pro, 
>etc. (on ia32 machines).
>I'm not an expert on this, though, so YMMV, but we've been running into this 
>ourselves under different circumstances.  Specifically, I don't have any 
>experience with >4GB IA32 boxes.
>On Monday 20 January 2003 12:23, Marc Schmitt wrote:
>> Hi all,
>> This is not very Dell specific, but I hope someone of you has
>> encountered this problem and knows what to do about it.
>> I'm running a small C++ program (attached) that does nothing else than
>> endlessly allocate chunks of 1MB of memory on various RedHat 7.3
>> installations (kernel 2.4.18-19.7.xsmp) on different Dell PowerEdge
>> servers. Those servers have between 2 and 4 GB RAM and at least 2GB of
>> SWAP. Having said that, I expect the program to be able to allocate
>> around 3.5GB of RAM (or whatever the per process maximum is on an IA32
>> machine).
>> But:
>> < --- zip --- >
>> Total allocated:2930 Mbytes
>> Allocation of 1 Mbytes
>> Total allocated:2931 Mbytes
>> Allocation of 1 Mbytes
>> Abort (core dumped)

Wes Perdue
IT Manager, Greenfield Networks

More information about the Linux-PowerEdge mailing list