From: marcf@nexus.yorku.ca (Marc G Fournier) Subject: Re: Question and Observations Date: Tue, 29 Dec 1992 00:10:58 GMT
junio@twinsun.com (Jun Hamano) writes:
>Marc G Fournier <marcf@nexus.yorku.ca> asks a couple of questions
>and comments in <marcf.725558783@yorku.ca>.
>As to passwd expiration problem, read manual page of `passwd' and
>find how to use `-x' option.
Will do :)
>About mkfs `-i'. It doesn't affect the blocksize. Traditional
>unix filesystem (and MINIX filesystem linux uses) allocates
>storage for inode area and blocks when the filesystem is
>initialized, and mkfs have to figure out how many inodes it needs
>to allocate for the filesystem being created. The ideal
>configuration is that you run out inodes and block storage space
>at the same time; otherwise either one of these would be wasted.
>If you expect you will have a lot of small files, then give
>smaller value to `-i' option so that you won't run out of inodes
>before you run out of disk space.
Okay, but one thing that I liked about the version I'm
using now is that you could stipulate different block sizes. I'm
currently using 512 byte blocks as opposed to the current Linux 1k
blocks. The benefit that I see with 512byte blocks is that ifyou have
a file that is only 499bytes (or something less then an even multiple
of 1k, you are only wasting 512bytes instead of 1k of the file system.
for example...
I create a fs of 70 Meg. Under linux, that works out to be
70,000 blocks, whereas under a 512byte block system, that works out to
be 140,000 blocks. therefore, I can hold 140,000 x 400byte articles
under the 512byte system as opposed to 70,000 x 400 byte articles under
Linux, thereby also wasting half the hard drive.
>About your `gripes' section; I too have the same feeling about
>SLS distribution, but this is certainly _not_ the problem with
>linux itself. We should send bug-reports and suggestions to
I do apologize for the obvious misconception that I put out :(
I think that Linux itself is great...the net is great at responding
to problems, Linus himself has developed a great kernel (from what
I know about kernels *smile*) and those that are 'creating' packages
are doing a great job, considering how fast things are growing and
being developed. *hat off to all involved*
>people (person?) at SoftLanding so that the next release will
>have more reasonable configuration. From what I've seen in
>c.o.l, about 30% of newbee questions and gripes wouldn't have
>been posted in the first place if SLS release did things right
>(e.g. "/etc/passwd not being readable by everybody" comes to
>mind). It is a trivial, but a tad tedious task to fix all the
>ownership and permission bits and such, but SLS release has so
>much influence to those naive linux users with little or no unix
>experience, we should urge and help SLS to provide more
>reasonable and easy to manage system configuration. I think it
>would benefit us, too; we wouldn't have to be bothered by trivial
>questions every time somebody finds out his users get error
>message from `who' command, for example.
the thing that I would honestly like to know is...from
peter mcdonald (I think)...have you ever set up the SLS distribution
under a multi-user environment? From the way I see things running,
and permissions set, I don't see how...unless what is being distributed
is not the same?
BTW, Peter...I do appreciate the distribution that you are
working hard at maintaining, and plan on sticking with it...I've
even dedicated an area on my file system for SLS for other ppl to
access who don't have ftp access.
Thanks...
Marc