Hello All:
Running 5.2 at our university. We have several student's processes that take up too much memory. Our system have 64G of RAM and some processes take close to 32-48G of RAM. This is causing many problems for others. I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it. Any thoughts or ideas?
TIA
Hi,
On Thu, Oct 16, 2008 at 12:48, Mag Gam magawake@gmail.com wrote:
I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it.
You can limit the amount of virtual memory of a process with "ulimit -v". See "help ulimit" or "man bash" for more details.
HTH, Filipe
Yes. Thanks. I was thinking of that too. Any other suggestions?
TIA
On Thu, Oct 16, 2008 at 12:56 PM, Filipe Brandenburger filbranden@gmail.com wrote:
Hi,
On Thu, Oct 16, 2008 at 12:48, Mag Gam magawake@gmail.com wrote:
I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it.
You can limit the amount of virtual memory of a process with "ulimit -v". See "help ulimit" or "man bash" for more details.
HTH, Filipe _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Thu, 16 Oct 2008 at 12:48pm, Mag Gam wrote
Running 5.2 at our university. We have several student's processes that take up too much memory. Our system have 64G of RAM and some processes take close to 32-48G of RAM. This is causing many problems for others. I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it. Any thoughts or ideas?
Have a look at /etc/security/limits.conf.
Mag Gam wrote:
Hello All:
Running 5.2 at our university. We have several student's processes that take up too much memory. Our system have 64G of RAM and some processes take close to 32-48G of RAM. This is causing many problems for others. I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it. Any thoughts or ideas?
In /etc/profile, use "ulimit -v NNNN" (in kilobytes) to limit the max virtual of all processes spawned by that shell
32G per process on a 64G machine sounds like a bit much. wouldn't a limit more like 4GB per user session be more appropriate on a multiuser system?
Hi John:
Well, we run a lot of statistical analysis and our code loads a lot of data into a vector for fast calculations. I am not sure how else to do these calculations fast without loading it into memory. Thats why we have to do it this way.
TIA
On Thu, Oct 16, 2008 at 1:00 PM, John R Pierce pierce@hogranch.com wrote:
Mag Gam wrote:
Hello All:
Running 5.2 at our university. We have several student's processes that take up too much memory. Our system have 64G of RAM and some processes take close to 32-48G of RAM. This is causing many problems for others. I was wondering if there is a way to restrict memory usage per process? If the process goes over 32G simply kill it. Any thoughts or ideas?
In /etc/profile, use "ulimit -v NNNN" (in kilobytes) to limit the max virtual of all processes spawned by that shell
32G per process on a 64G machine sounds like a bit much. wouldn't a limit more like 4GB per user session be more appropriate on a multiuser system? _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Mag Gam wrote:
Hi John:
Well, we run a lot of statistical analysis and our code loads a lot of data into a vector for fast calculations. I am not sure how else to do these calculations fast without loading it into memory. Thats why we have to do it this way.
well, if you got several processes that each need 32GB in a 64GB machine, you're gonna end up swapping.
the traditional way of doing this sort of thing on limited memory machines was to take a sequential pass through the data, calculating the statistics on the fly. I know that kind of thing is very difficult for some algorithms (FFT's are notorious for being unfriendly to sequential processing), but for many algorithms, a few sequential passes of calculations can be /faster/ than random access and swapping when theres memory/process contention.
2008/10/17 Mag Gam magawake@gmail.com:
Hi John:
Well, we run a lot of statistical analysis and our code loads a lot of data into a vector for fast calculations. I am not sure how else to do these calculations fast without loading it into memory. Thats why we have to do it this way.
About 15 years ago I changed an application on SGI IRIX from using text files scanf(3)'ed into memory (with floating point numbers in them) to binary files mmap(2)'ed into memory. Processing time was cut down by over 95% and did much more in the 5% left (e.g. allow interactive real-time viewing of different "frames" of data).
Using mmap'ed files means that the system will know that these pages are backed by blocks on the file system and therefore it won't take up so much "buffer" space which needs to be writen out into the swap partition whenever the memory buffer is needed for something else, only disk cache space which can be just freed if the buffer was only read. You can also benefit if multiple processes access same file - they'll share the buffer in memory too.
It's not a silver bullet, there are still issues with too random access causing the system the thrash, but at least it won't take up so much swappable memory, it'll save lots of copying (file->kernel->user when reading and the other way around when writing), system calls etc.
If you can process data in sequential order and possibly with help of madvise(2) you can probably squeeze out even more from this option.
--Amos
ulimit is good for per process. What about for total usage? If a user has 5 processing -- each taking up 10G, will account for 50G. Is there a way to avoid this? Or have the VM be sensative, once its swapping we want to start killing the processing that take the most memory?
TIA
On Wed, Oct 22, 2008 at 1:58 AM, Amos Shapira amos.shapira@gmail.com wrote:
2008/10/17 Mag Gam magawake@gmail.com:
Hi John:
Well, we run a lot of statistical analysis and our code loads a lot of data into a vector for fast calculations. I am not sure how else to do these calculations fast without loading it into memory. Thats why we have to do it this way.
About 15 years ago I changed an application on SGI IRIX from using text files scanf(3)'ed into memory (with floating point numbers in them) to binary files mmap(2)'ed into memory. Processing time was cut down by over 95% and did much more in the 5% left (e.g. allow interactive real-time viewing of different "frames" of data).
Using mmap'ed files means that the system will know that these pages are backed by blocks on the file system and therefore it won't take up so much "buffer" space which needs to be writen out into the swap partition whenever the memory buffer is needed for something else, only disk cache space which can be just freed if the buffer was only read. You can also benefit if multiple processes access same file - they'll share the buffer in memory too.
It's not a silver bullet, there are still issues with too random access causing the system the thrash, but at least it won't take up so much swappable memory, it'll save lots of copying (file->kernel->user when reading and the other way around when writing), system calls etc.
If you can process data in sequential order and possibly with help of madvise(2) you can probably squeeze out even more from this option.
--Amos _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Amos Shapira -
Filipe Brandenburger -
John R Pierce -
Joshua Baker-LePain -
Mag Gam