Re: Next openSUSE

On Wed, 26 Aug 2009 18:31:58 +0100, David Bolt <blacklist-me@xxxxxxxxxx> wrote:

Many problems with ourdinary users are caused by fear to
touch the computer ... "something might go wrong". Some persons even get
hysterical about it.

That's where children come in. They don't have seem to have that fear.

Exactly. The point with children: They do try. They are not afraid of mistakes.
They do not give up. THIS one has to tell the adults.

The adults always compare themselves with children and believe that the children
are smarter. But the children aren't smarter, they simply go on trying and they
do not count their tries.

Once you got the adults to realize this, teaching them is a lot easier.


Installing an OS and maintaining it is NOT something for the standard
user. I do not maintain my car. The only things I do myself is wash it,
put gasoline in it, water for the windows and air for the tires. For all
the rest (including changing and adding oil) I go to somebody who knows
what he is doing.

You say it. Tell someone you know how to install a problem. He looks at
^^^^^^^
you if you are an idiot.

I can understand that :-) Now, if you really meant program...

Heck, those typos. I just made a mistake. Somehow the mix keybord:coffee was
wrong. I guess next time I do not take 1 cup of coffee...

So, this is a new keyboard with quite a different touch feeling... I'll see to
which typos that leads... ;O)


Why would I then expect that everybody should be able to administer its
computer. They can't.

Well, this is what help-desks ar for... ;O)

Hahahahaha. I gather you've actually never spoken to many help desks? There
are large numbers of comments made by those that have tried to use them that
the help desk droids are frequently clueless script-readers.

Not many. IBM doesn't have many. We were special customers, so we talked with
the techies who came to our offices, and with the ?level support. Higher the
hierarchy did not go for us. But at least SOME papers were helpfull.


For somebody who has to maintain systemes with a high amount of files,
you seem to know very little about it. Just one question, what is the
best filesystem for a huge number of files and why?

I do not know what is best for a large number of files. I have to take
what I can handle easily at least cost.

I've found that ext2/3 isn't a bad general purpose file system, although
ext3 is preferred because it uses a journal which aids in recoveries when
there's an unforeseen, and unplanned-for, shutdown.

I do not know of other systems. I had to take what was delivered by the
installer.


For large numbers of small files, reiserfs seems to be better. I've used it
where there have been over 100,000 files in a single directory. For very
large files, maybe of DVD-R sizes and larger, I prefer to use xfs. There are
others, but you'd have to experiment with them to find out just which ones
handled things the best. Or just read up and see which ones other people
prefer, and why.

reiserfs was a terrible experience. I do not know if it was the reiserfs or oner
or more "unfit" parameters, but ext3 seems to be less error-prone.


I was told that Linux would be excellent - so I tried it. The first
disappointment was to see that it handles less files in a directory than
DOS.

I don't know the directory limits of the DOS file system, but I'd be
surprised if it was higher than any of the other file systems Linux
supports.

I did not go to the top, but > 16k is possible. So my guess is that it might be
32767 or 32766...

For the time being, that was terrific.


When I tell them something about this or that on my machines, they get
eyes like saucers.

Show us something, please.

Like downloading 10 million files in a row.

10 million files sounds like a bit of an exaggeration, unless you're talking
about files that have an average size a lot less than 1KiB. If not, you'd be
talking about downloading upwards of 9GiB.

Yes. I forgot the size. Some were small (I forgot the minimum file size) and
some could be 40 or more kBytes. Javascript spoils a lot...

So I had to weed out the unneccessary stuff, and had to make pauses for these
things. It took several weeks...


Like having 100,000 file sin a
directory.

Been there, done that, changed the file system for that partition because I
was using ext3 at the time and it was damned slow when trying to list the
directory contents or delete files.

Delete? Whenever I tried to delete more than 3000, the thing would crash. Trying
to delete from the command line gave such fine errors like "to many files". The
exact wording I forgot. With DOS that could have been some key-clicks...


Like having an archive of (zipped) 200 MegaBytes concerning a
certain MLM-scheme.

I don't have any MLM schemes zipped up, but I do have archives of 200MiB and
bigger. One I very recently deleted contained an old copy of my website that
dated back to when I was providing packages for SuSE 9.3 and earlier. Since
I have no reason to keep packages for versions that old, there was no reason
to keep the archive.

We prepared those ZIPs for the attorneys of state. ;O)


Like working on text files with 15,000 lines.

I work on a bash script that has a few thousand lines, without having any
issues. Okay, I don't have any issues working with the script. I may
occasionally _cause_ issues by breaking it a little, but for the most part
it works just fine.

I work on other text files with thousands of lines. None of the programs I
use have issues with them, and I certainly don't think there's anything
strange about it.

Large programs can be so long. In my cases it is collected material which is to
be prepared for further investigations or to be given to attorneys of state or
end up as large web pages.


That proxy I need for something else: to hide my IP.

So use TOR. It's set up specifically so you can use the net without giving
out your IP address.

Too fast switching - and they throw you out. They are very strict with that...


The problem is not with me, but mostly with people who receive links by
email, etc. Some folks we investigate are extremely sharp in looking at
their server reference log...

There are no referrers passed on when clicking on a link in a mail program.
This I've verified by looking at links I've posted in both emails and usenet
postings, and the referrer is always empty. However, if you're using a web-
mail provider or Google to read newsgroups, that is a completely different
matter.

I do not know which browser you use. Mine I set to refferer=OFF. The problem
with the defaults is that referrer and other mess are set ON.

Some email providers servers do NOT use the actual link, but have an
intermediate one, so cut+paste will not work. I do not know why they are so sick
for links (because these links ARE in the email they transport), but for
ordinary users such things may have bad consequences...

If you have a web-site, then just look into the referrer log. If not you, so
then someone else will now or then leave such traces.

--
Brettl, die eine Welt bedeuten

http://www.ariplex.com/ama/ama_ml39.htm
.

Relevant Pages

  • Re: [patch] ext2/3: document conditions when reliable operation is possible
    ... wouldn't it be an ext3 bug? ... many naive filesystem designers make is that writes succeed or they ... This will obviously cause a write failure, ... No file system will recover that data although you might be able to scrape out some remaining useful bits and bytes. ...
    (Linux-Kernel)
  • Re: How do you get to the DOS prompt using the Recovery CD
    ... NT has no DOS mode, so you'd have to run that from a DOS mode boot ... Windows writes to the HD. ... then repairing internal file system logic so data can be evacuated. ...
    (microsoft.public.windowsxp.configuration_manage)
  • Re: want to speed up laptop
    ... Nearly hit all points on my laptop, ext3 plus ... But I'm not good at tuning the file system. ... There are several factors which combine to hurt overall disk ... Assuming the laptop is reliable, ...
    (Debian-User)
  • Re: Next openSUSE
    ... Heck, those typos. ... Google them. ... I've not had any major issues with any file system, ... problem with the defaults is that referrer and other mess are set ON. ...
    (alt.os.linux.suse)
  • Re: Slow ruby regexes
    ... I don't know about proposed plans for the regexp engine for ruby, ... with one of our sys admins who wanted to use the Riser FS rather than Ext3 as ... the file system on a new server. ... was that file system performance was not a significant bottleneck for the ...
    (comp.lang.ruby)