On Tue, Aug 09, 2005 at 12:08:23PM +1000, David Chinner wrote:
> Swap won't help if you're running an ia32 (32bit) kernel - you have a
> per-process memory limit of 1-4GiB (depending on kernel and config).
> The amount of physical memory and swap does not change this
> limitation.
I have an ia32 box running Debian Sarge (xfsprogs 2.6.20) with Linux
2.6.12.4, 3GB physical memory, and at present 2GB swap. We don't have a
64-bit system with more RAM to which we can transfer our filesystems, if
ever, so I'd like to know:
1. What kernel configuration do we need to enable in particular to
maximize the per-process memory limit?
2. What limit should I follow for per-filesystem size, so that I can be
confident that even when the filesystem has reasonably many inodes both
xfs_check and xfs_repair will work without hitting the per-process
memory limit of Linux on ia32?
Also, I am using LVM2 (2.01.04) to allow me to build my XFS filesystems
across multiple physical volumes (each physical volume is an external
hardware RAID 5 enclosure connected to the system via SCSI). I wonder:
1. Are there any known issues with XFS and LVM2 when the total volume
group size gets very big (eg: >2TiB)? Or does XFS not care and will this
be a purely LVM issue?
2. Aside from problems with snapshots, are there any known issues with
running XFS on top of LVM2?
Cheers!
--> Jijo
--
Federico Sevilla III : jijo.free.net.ph : When we speak of free software
GNU/Linux Specialist : GnuPG 0x93B746BE : we refer to freedom, not price.
|