Mailing List ArchiveSupport open source code!
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]Re: tlug: LILO Vs. 1024??
- To: TLUG mailing list <tlug@example.com>
- Subject: Re: tlug: LILO Vs. 1024??
- From: Rex Walters <rex@example.com>
- Date: Wed, 7 Oct 1998 14:32:30 +0900
- Content-Transfer-Encoding: 7bit
- Content-Type: text/plain; charset=us-ascii
- In-Reply-To: <Pine.LNX.3.96LJ1.1b7.981007125730.1878D-100000@example.com>; from Jonathan Byrne - 3Web on Wed, Oct 07, 1998 at 01:02:36PM +0900
- Mail-Followup-To: TLUG mailing list <tlug@example.com>
- References: <19981007111157.E11845@example.com> <Pine.LNX.3.96LJ1.1b7.981007125730.1878D-100000@example.com>
- Reply-To: tlug@example.com
- Sender: owner-tlug@example.com
On Wed, Oct 07, 1998 at 01:02:36PM +0900, Jonathan Byrne - 3Web wrote: > On Wed, 7 Oct 1998, Rex Walters wrote: > > >I *strongly* recommend by the way that you partition this thing into > >multiple partitions. I don't believe in one big ext2 filesystem for > >all of your Linux data. > > What benefits does this get me, and/or risks if I don't? A filesystem is just a very complex data structure. The leaf nodes in the data structure contain your data. Losing or corrupting data in a leaf node is bad, but losing or corrupting *metadata* (info about the structure itself) can be catastrophic. The main reason to use multiple filesystems (IMHO) is to compartmentalize this risk. Corrupting (or simply filling) filesystem A won't affect filesystem B in any way. At the very least, you almost certainly want to keep *your* data in separate filesystems from OS data. You'll appreciate this the first time you try to upgrade your operating system (or if you want to try multibooting different OS partitions). It's nice to be able to have separate / & /usr partitions for TL/RH/Debian or say for a 2.1 kernel and 2.0 kernel and still be able to mount the same home directory under all of them. And having the install scripts for your upgrade newfs the filesystem containing your home directory is never fun. You also gain administrative flexibility. You can choose different optimizations for each filesystem (space/performance, number of inodes, max-mount-counts before fsck, raid choices, etc.). You can optimize based on data flow and access patterns. Read-only filesystems can be configured differently from heavy-write filesystems, mostly-sequential from mostly-random, mostly-logs from mostly databases, and so on. You're experiences recently should point out the benefits of a small root partition wrt BIOS limitations. The ability to do "full dump" image backups and restores is also a big plus. This is especially true if you're able to mount a filesystem read-only -- you *know* you can safely restore a dd image if it was created while the filesystem was read-only. There are, I'm sure, even more reasons that I've missed, but fundamentally the reason to have multiple filesystems is that not all of your data is the same. I actually stumbled across another benefit of multiple filesystems at home the other day. I've actually got several large disks on my system at home, each chunked up into multiple partitions. I'd created filesystems on all of these partitions all on the same day. Since I don't leave my home machine running all the time, all of these filesystems are mounted and unmounted pretty frequently (at least once or twice a day). Linux as a sanity check forces a fsck on a filesystem at boot time even if it was cleanly unmounted if a certain amount of time has elapsed or too many unmount/mount cycles have occured (20 by default). Every week or two I'd go to boot my machine and have to wait an infuriatingly long time for everything to come up as multiple 2+ GB partitions were fsck'd (even with fstab saying to do 'em in parallel it took a while). The solution was to run "tune2fs -c" on each filesystem and specify a different max-count before check for each. This way I should never have to fsck more than one or two filesystems during a single boot. I also took the opportunity to bump the frequency of forced fscks way down. Much nicer. Convinced? Regards, -- Rex --------------------------------------------------------------- Next Meeting: 10 October, 12:30 Tokyo Station Yaesu central gate Featuring the IMASY Eng. Team on "IPv6 - The Next Generation IP" Next Nomikai: 20 November, 19:30 Tengu TokyoEkiMae 03-3275-3691 --------------------------------------------------------------- Sponsor: PHT, makers of TurboLinux http://www.pht.co.jp
- Follow-Ups:
- Re: tlug: LILO Vs. 1024??
- From: Jonathan Byrne - 3Web <jq@example.com>
- References:
- Re: tlug: LILO Vs. 1024??
- From: Rex Walters <rex@example.com>
- Re: tlug: LILO Vs. 1024??
- From: Jonathan Byrne - 3Web <jq@example.com>
Home | Main Index | Thread Index
- Prev by Date: Re: tlug: Downward Spiral
- Next by Date: Re: tlug: Wine and Windows Printers
- Prev by thread: Re: tlug: LILO Vs. 1024??
- Next by thread: Re: tlug: LILO Vs. 1024??
- Index(es):
Home Page Mailing List Linux and Japan TLUG Members Links