|
|
| Author |
Message |
| 20 new of 69 responses total. |
remmers
|
|
response 50 of 69:
| 
Nov 9 13:56 UTC 2001 |
Too much resource overhead on the current hardware. We should
be able to when we move to new hardware.
|
jp2
|
|
response 51 of 69:
|
Nov 9 14:58 UTC 2001 |
This response has been erased.
|
steve
|
|
response 52 of 69:
|
Nov 9 21:04 UTC 2001 |
No, it doesn't. It's rock solid. It does its job very very well,
and because of its stability and staff's efforts we've not had serious
security breaches on quite some time.
SunOS is old however, and shows signs of that here and there. Disk
quotas are one example. There are others, which is why we'll be moving
to OpenBSD at some point. But have respect for SunOS: its one hell of
a good OS. It just isn't modern anymore.
|
other
|
|
response 53 of 69:
|
Nov 9 21:26 UTC 2001 |
Not that I want to side with Jamie or anything like that, Steve, but in
HIS world, SunOS does blow goats.
Whether or not that is the case in anyone else's world is not the
question at hand, and disputing him on the point will only aggravate you.
Don't waste your effort explaining the values of a good thing to someone
who only knows from bad things.
|
cross
|
|
response 54 of 69:
|
Nov 9 21:56 UTC 2001 |
Actually, SunOS does suck. Reliability shouldn't be the sole metric
by which an OS is judged; we should also consider design elegance,
performance, adaptability to different tasks, and etc. See, for an
introduction to some of the problems in SunOS's design, the following
paper: http://www.caldo.demon.co.uk/doc/taste.pdf. Granted it's
somewhat dated, but many of the problems pointed out are fundamental
and never really got solved. Perhaps equally amazing is how the `Unix'
community carries forward these mistakes like some sort of red-badge-of-
courage-esque symbol of commitment to their roots. Funny for a system
whose malleability and reaction to change was its greatest strength.
|
mdw
|
|
response 55 of 69:
|
Nov 9 22:37 UTC 2001 |
The problems he describes aren't unique to, or even particularly
associated with SunOS. Virtually the identical complaints could have
been made about Linux. Perhaps the real culprits are the Japanese; by
making memory so cheap, they've encouraged sloppy programming.
|
cross
|
|
response 56 of 69:
|
Nov 9 23:00 UTC 2001 |
The problems are, as you say, indemic to Unix variants. I should note
that I also happen to think that Linux not only sucks, but *really*
sucks; hence my comments about carrying forward the legacy and so on.
I think you should blame Shannon.
|
keesan
|
|
response 57 of 69:
|
Nov 9 23:32 UTC 2001 |
Is there a big difference between allowing only paying members to post images
on grex, and allowing only paying members outgoing ftp access which you need
to post images at a free website from grex? Is this not equally
discriminatory?
|
scott
|
|
response 58 of 69:
|
Nov 10 00:23 UTC 2001 |
Not really. You could always find another way other than Grex to ftp to a
free image site.
|
scott
|
|
response 59 of 69:
|
Nov 10 00:25 UTC 2001 |
Oops, I didn't read the question very carefully.
I don't think Grex would end up allowing images only for paid members. Most
of us don't like the idea/precedent of being able to pay for better services.
The current ftp policy is based on the identifiable user issue than any money
issue.
|
keesan
|
|
response 60 of 69:
|
Nov 10 02:27 UTC 2001 |
But this means paying users can use grex to post images to a free website,
and nonpaying users cannot post images without going to a library or somewhere
else. So they have less services. Not complaining, just pointing out that
the result is the same.
|
janc
|
|
response 61 of 69:
|
Nov 10 04:16 UTC 2001 |
(Actually, I've never been able to find any evidence that quotas don't work
OK under SunOS, except for some references to some bugs on Sun 3's that were
fixed by patches long ago. If we weren't moving off SunOS, I'd be doing a
round of experiments to see if SunOS quotas could work for us. But since we
are leaving SunOS, the point is moot.)
|
gelinas
|
|
response 62 of 69:
|
Nov 10 04:37 UTC 2001 |
(Quotas have been used on the SunOS 4.1.4 machines I've used. Other than
being per-filesystem, they worked as advertised, as far as I could tell
as an average user.)
|
malymi
|
|
response 63 of 69:
|
Nov 10 10:38 UTC 2001 |
re #19: porn sites also use hop-off (or jump-through) sites, which only
have small images and some html, which takes you to the real site.
so allowing any images is likely to attract them.
this isn't to say that i have an opinion on the issue, one way or
the other.
|
malymi
|
|
response 64 of 69:
|
Nov 10 10:45 UTC 2001 |
re #22: there is no way for a web server to tell the difference between
a page that displays an image and one that requires you to click
on a link to obtain the image. it should be possible to write
a filter to insure that served html only contains href's and
no img's.
|
malymi
|
|
response 65 of 69:
|
Nov 10 11:04 UTC 2001 |
re #38: in order to use geocities to provide hosting of images without
violating their aup (acceptible use policy) you would have to
use an html file, hence it cannot be used as the target of an
img src tag in a page hosted at grex.
|
remmers
|
|
response 66 of 69:
|
Nov 10 18:17 UTC 2001 |
Re resp:57 - There's a significant difference in bandwidth cost
between (1) a member using ftp on grex to upload an image to a
remote website and (2) the member storing the image in their
grex www directory. In case (1), the image uses grex bandwidth
only once, when it's uploaded by the owner to the remote site.
In case (2), the image uses grex bandwidth every time anybody
on the web views it.
|
other
|
|
response 67 of 69:
|
Nov 10 19:18 UTC 2001 |
re #65:
<html><body>
<!-- this file = "http://annoyingserver.com/mydirectory/image01.html" -->
<img src="image01.gif">
</body></html
|
steve
|
|
response 68 of 69:
|
Nov 10 23:54 UTC 2001 |
Quotas do *not* work on a heavily used SunOS 4.1.4 system. Sure, the
typical
little used system is likely fune with them, but not when the filesystem is in
constant use. I've gotten that advice several times now from people who used
to use SunOS, years ago.
|
gelinas
|
|
response 69 of 69:
|
Nov 12 19:36 UTC 2001 |
I asked a fairly knowledgeable sysadmin aboout SunOS quotas. The problem is
that the quota table is a fixed size; if you have more users than the table
will hold, the 'extra' users don't get quotas. I'm told that in SunOS 4 it
was possible to use adb to resize that table and then rebuild the kernal,
while SunOS 5 has a variable that can be used to resize it. Apparently,
the table takes up the same amount of memory no matter how fully populated,
so systems with a really large userbase are going to take a performance hit.
So quotas work, but they may not be worth the trouble. I think that's
where we are right now.
|