On Sun, Nov 27, 2011 at 01:22, Mick <michaelkintz...@gmail.com> wrote: > On Friday 25 Nov 2011 20:08:01 Walter Dnes wrote: >> On Fri, Nov 25, 2011 at 11:12:42PM +0700, Pandu Poluan wrote >> >> > Everytime I read some guide on LVM, my eyes becomes blurry, the room >> > starts spinning, and I can hear wolves howling ... :D >> > >> > Seriously, LVM looks mighty nice, but it also looks (and is!) mighty >> > complex. >> > >> > So, I want to start from something simple. >> > >> > Aaaanyways, after reviewing my production boxes, I decided to >> > implement the following strategy: >> > >> > / == 800 MiB >> > /boot == 20 MiB >> > /usr == 1800 MiB >> > /usr/portage == 2000 MiB >> > /var == 4000 MiB >> > /var/lib/postgresql == 1000 MiB >> > >> > Comments, suggestions, are welcome :) >> >> I have my own weird approach that's even weirder than my mdev setup<G>. >> I start with... >> * 250 megabytes for / as ext2fs (No that is not a typo) >> * 4 gigs for swap >> * the rest of the drive is /home as one huge reiserfs partition >> >> And I do *NOT* use LVM. "fdisk -l" shows... >> >> Device Boot Start End Blocks Id System >> /dev/sda1 2048 976773167 488385560 5 Extended >> /dev/sda5 4096 516095 256000 83 Linux >> /dev/sda6 518144 8906751 4194304 83 Linux >> /dev/sda7 8908800 976773167 483932184 83 Linux >> >> "df" shows >> >> Filesystem 1K-blocks Used Available Use% Mounted on >> rootfs 247919 29315 205804 13% / >> /dev/root 247919 29315 205804 13% / >> devtmpfs 10240 0 10240 0% /dev >> rc-svcdir 1024 44 980 5% /lib/rc/init.d >> mdev 10240 0 10240 0% /dev >> shm 1551308 0 1551308 0% /dev/shm >> /dev/sda7 483917384 251951296 231966088 53% /home >> >> The secret is that I bindmount /opt, /var, /usr, and /tmp onto the >> large reiserfs partition. >> > ########################################################################### >> ### /dev/sda5 / ext2 noatime,nodiratime,async >> 0 1 /dev/sda7 /home reiserfs >> noatime,nodiratime,async,notail 0 1 /home/bindmounts/opt /opt auto >> bind 0 0 /home/bindmounts/var /var >> auto bind 0 0 /home/bindmounts/usr /usr >> auto bind 0 0 /home/bindmounts/tmp >> /tmp auto bind 0 0 /dev/sda6 >> none swap sw 0 0 /dev/cdrom >> /mnt/cdrom iso9660 noauto,users,ro 0 0 /dev/cdrom1 >> /mnt/cdrom1 auto noauto,user,ro 0 0 /dev/sdb1 >> /mnt/extb auto noauto,user,noatime,async 0 0 /dev/sdc1 >> /mnt/extc auto noauto,user,noatime,async 0 0 >> >> # glibc 2.2 and above expects tmpfs to be mounted at /dev/shm for >> # POSIX shared memory (shm_open, shm_unlink). >> # (tmpfs is a dynamically expandable/shrinkable ramdisk, and will >> # use almost no memory if not populated with files) >> shm /dev/shm tmpfs nodev,nosuid,noexec 0 0 >> > ########################################################################### >> ### > > I recall your interesting mounting approach, but never really understood the > benefit of it. Would you please explain why you use bindmount? >
Not really explaining waltdnes' interesting layout, but using bindmount (instead of symlinks) ensures that when a program tries to find a relative directory from a path, it will not attempt to do so from the symlink's target. E.g.: Say I have /lib/gzampl/, which is actually a symlink/bindmount to /mnt/gzampl/. Then theres another directory /lib/morethings. With a symlink, if a program wants to do "../morethings" from within /lib/gzampl/, it might end up in /mnt/morethings if the program tries to resolve the symlink first. With a bindmount, doing "../morethings" from /lib/gzampl/ will always end up in /lib/morethings. (CMIIW) That said... mentioning bindmount made me rethink things... What if I have: /mnt/.temporaries ==> ext4, 4GiB /mnt/.persistents ==> reiserfs, 2GiB then I make some directories and bindmounts: /mnt/.temporaries/tmp --bm--> /tmp /mnt/.temporaries/vartmp --bm--> /var/tmp /mnt/.temporaries/run --bm--> /run /run --bm--> /var/run /run/lock --bm--> /var/lock /mnt/.persistents/postgresql --bm--> /var/lib/postgresql /mnt/.persistents/vardb --bm--> /var/db /mnt/.persistents/varlog --bm--> /var/log /mnt/.persistents/varspool --bm--> /var/spool Ta da! The ephemeral directories can now just fight among themselves, and the important directories can be backed up in one fell swoop (via /mnt/.persistents)? Thoughts are welcome, of course :) Rgds, -- FdS Pandu E Poluan ~ IT Optimizer ~ • LOPSA Member #15248 • Blog : http://pepoluan.tumblr.com • Linked-In : http://id.linkedin.com/in/pepoluan
