Messages in this thread | | | From | "Khimenko Victor" <> | Date | Thu, 18 Feb 1999 23:49:05 +0300 (MSK) | Subject | Re: 2.2.1 mkfs.ext2 out of memory |
| |
In <Pine.LNX.3.95.990218143725.180A-100000@chaos.analogic.com> Richard B. Johnson (root@chaos.analogic.com) wrote: > On Thu, 18 Feb 1999, Jon P. deOng wrote:
>> Hi, >> I'm running a pII 400 with 256Mb of ram with a vanilla 2.2.1 kernel >> Im trying to create a raid5 array with 4 9gig scsi seagate cheetah drives >> connected to the external adapter of an adaptec 2940uw controller. My os >> drive is also a 9gig scsi cheetah on the internal ultra scsi adapter. I >> have no scsi2 devices at all. I have verified proper termination and turned >> off the auto term in the adaptec bios. >> Now I have no problems with mkraid, mdadd, or mdrun. All of these raid tools >> report no errors. The problem comes when i try to format /dev/md0 >> my syntax is >> mkfs.ext2 /dev/md0 >> mkfs gets almost halfway through writing the inode tables and then i get >> this output >> >> Writing inode tabels: 1969/3254 >> Out of Memory for mkfs.ext2 >> Killed
> Yep! First make a big swap partition or a big swap file. There is, roughly > 4096 bytes of memory required for every megabyte of file-system space when > you execute mkfs.ext2. This depends upon the allocation-unit size, but > it's a reasonable rule-of-thumb.
> Then boot into single-user mode with no network daemons running. Make > sure your swap partition is activated, then make the file-system. You > should be sucessful.
Sometimes big swap helps but NOT ALWAYS. I was unable to set up RedHat 4.2 on comp with 11MiB RAM and 1GiB HDD even with 128MiB swap but with 31MiB all was just fine without any swap :-(( Plus your "rule-of-thumb" will give us 144MiB for 36GiB "drive" and there are 256MiB REAL memory ! IMO it's some rare race in kernel triggered by mkfs.ext2 when there are a lot of dirty pages in cache but they are not written on disk and so you could not allocate memory even of there are plenty of swap area...
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |