wjones@nwnexus.WA.COM (Warren Jones) (01/14/91)
We have a single hard disk on our system (an RS/6000 under AIX 3.1). Can anyone explain the rational for dividing this disk into several partitions, and then re-joining them (via mounts) into a single directory structure? I believe that earlier versions of Unix had 16 bit inode numbers, resulting in a limit of 64K files/file system. Thus if you wanted more than 64K files, you would have to mount multiple partitions. But AIX 3.1 has 32 bit inode numbers, so this rational does not apply. (I suspect 32 bit inode numbers are the norm for current Unix implementations -- how about your system?) As long as there is only a single physical disk, it seems more efficient to configure it as a single partition. This produces a single pool of free space. Otherwise, you could run out of space (for example) in /tmp on the root partition, even though there was still plenty of space in /usr/tmp (on the /usr partition). Am I missing something? Comments, advice and explanations are welcome before I take the major step of re-configuring our disk.
rickert@mp.cs.niu.edu (Neil Rickert) (01/14/91)
In article <457@nwnexus.WA.COM> wjones@nwnexus.WA.COM (Warren Jones) writes: >We have a single hard disk on our system (an RS/6000 under AIX 3.1). >Can anyone explain the rational for dividing this disk into several >partitions, and then re-joining them (via mounts) into a single >directory structure? As soon as someone accidently (or deliberately) runs a looping program that completely fills a partition you will discover the benefits of multiple partitions. At the very least keep a partition with essential system stuff, and space for log messages, which is separate from the partitions containing user directories. -- =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= Neil W. Rickert, Computer Science <rickert@cs.niu.edu> Northern Illinois Univ. DeKalb, IL 60115 +1-815-753-6940
kevin@msa3b.UUCP (Kevin P. Kleinfelter) (01/17/91)
rickert@mp.cs.niu.edu (Neil Rickert) writes: >In article <457@nwnexus.WA.COM> wjones@nwnexus.WA.COM (Warren Jones) writes: >>We have a single hard disk on our system (an RS/6000 under AIX 3.1). >>Can anyone explain the rational for dividing this disk into several >>partitions, and then re-joining them (via mounts) into a single >>directory structure? > As soon as someone accidently (or deliberately) runs a looping program that >completely fills a partition you will discover the benefits of multiple >partitions. ... Once in a while, after a power-outage or crash, fsck may not like a few inodes here and there. :-) If you have a lot of activity on a partition, then that partition is most likely to have problems. I have observed fsck delete an awful lot of files from a /usr filesystem because of problems caused by activity to /usr/adm/errlog (on a SYS V machine). You can bet that I'm now a real believer in /usr/adm having its own partition, to protect other partitions. -- Kevin Kleinfelter @ Dun and Bradstreet Software, Inc (404) 239-2347 {emory,gatech}!nanovx!msa3b!kevin Soon to become {emory,gatech}!nanovx!dbses0!kevin (But not yet!)