Importance of Uptime

Importance of Uptime

Post by Mike » Sat, 16 Mar 2002 13:02:29






>>If even five minutes for a reboot is five minutes too much -- you need
>>Linux.  

> When you examine all the machines in a net block hosting the
> distributors home page, Windows 2000 leads by almost 100 days.  In
> fact the Microsoft net block has 15 machines with average uptimes
> better than the best performed machines in the net blocks hosting the
> Debian, Suse and RedHat Linux home pages.

So?  That may just show that the linux versions update their hardware
and/or basic parts of the OS (like the kernel) more often than sites
hosted in Microsoft's netblock.  I wonder how any major flaws have
been fixed in Win2000 in the last year that the nearly 400 day uptime
indicates the people running the site hasn't botherd patching.

Why such a narrow focus, anyway?  Surely it'd be more interesting to
see the highest uptimes found anywhere, rather than trying to draw
conclusions of how long systems run for at particular places that
may have completely different operating practices?

  http://uptime.netcraft.com/up/today/top.avg.html

Interesting reading, right?

Try again.
--
Mike.

 
 
 

Importance of Uptime

Post by Albert Solson » Sat, 16 Mar 2002 14:11:13


Quote:> What you recall and what I could find by looking at the actual bug
> reports seem to be contradictory as well. Would you be so kind as to
> back that claim up?

   Just download code and test it on a 2.2.x kernel if you have got one,
I think it would work, but I am not sure after all ... it is long time ago.

Quote:> they call it "version 1.0".

   So kernel 2.4.0 should have been okay after all, shouldn't it ?

Quote:> you develop any software yourself at all?

   Pointless. Anyway, as a programmer, I am not a professional,
just hobbyst.

Quote:> > 2) Don't use 1.0 versions because they are not good.
> I never said that. The current kernel is, in fact, version 2.4.

   I understand you, but you should understand me too.

Quote:> Have you even read the kernel howto, btw?

   Too many personal questions. This is not a test about
my Linux knowledge, but yes, I have read it some times
since VM86 patch in the 2.0.x series. I just would like
that was then, in that 2.x.9 when it was really released,
can't I have personal preferences ?

Quote:> No. I don't intentionally run code I don't know the purpose for.

   That's right.

Quote:> Many
> peple do tho, and since many people run windows as admin since this
> is the default, they can cause much damage to their systems.

   The same that would run Linux as root if they used it in my personal
opinion, and I even think that this would be even more dangerous.

Quote:> I am the
> local admin on my windows box as well, hence I'm quite wary of what I
> click on.

   So they are a lot of persons. Running Windows is not insecure by
definition, you just have to be careful, just a little more than with
Linux. Anyway, NT, 2K and XP have different users as I think that
yo do know.

Quote:> You were giving an example of something that has been since fixed,
> claiming that many people run the kernel version that it affected.

   No, that possible many people. And I did not meant it was a
nowadays issue, just pointed an historical example (required by Kenneth)
of Linux crashed by non root user. Meant nothing else.

Quote:> You
> provided no factual evidence and urged other people to look it up for
> them selves in stead.

   If you want. It was not too hard, wan't it ? I told you to look for it
if you did not believe me, just that. My own post explainned it more
than less okay.

Quote:> That is not advocacy, you are not on topic.

   I was answering a direct question of Kenneth. And thanks for your
help, but being or not being on topic is my own decision while this
newsgroup is not moderated. By the way, I think I as on topic.

 
 
 

Importance of Uptime

Post by Albert Solson » Sat, 16 Mar 2002 14:18:35


   Sincerously, for my home desktop I do not need 100%,
and I am honest. Windows 2000 and Windows XP for
my home, give me the same uptime than Linux, since I
just have it about 10 hours a day.

   By the way, I honestly think that rebooting a machine
is good if it can be done without disturbing its services.
Why that uptime importance ? The Solaris and Linux
servers where I work are booted each weekend and
they still give us 100% time.

   I think 100% is really needed only in real time
environments.

 
 
 

Importance of Uptime

Post by mlw » Sat, 16 Mar 2002 14:41:44





> > Top post (sorry)

> > I like the spirit of this post. It is what advocacy is all about. And
> thank you
> > for making a case for "uptime." You put into words what we have always
> thought
> > but never taken the time to express.

> > An OS that askes you to reboot is one in which the designers failed. I can
> > understand upgrading to a different version, I can understand adding new
> > hardware, but come on now, adding software?

> > Nope, I started Windows development on version 1.03, the rebooot *was
> OK
> > when you had a floppy based system.

> There is never a reason why you *HAVE* to reboot Windows (except for service
> packs which replace the kernel and certain always in-use files), it's just
> that rebooting is easier than manually fixing things up.

Define "have" to reboot. When a dialog box says "Restart Windows now?" and only
gives you the option of "OK" I would say that is "have to reboot."

Quote:

> The usual reason to force a reboot is that some files are in-use when you
> install them.  You can minimize or even completely eliminate this problem by
> shutting down any services which depend on those files.  This will make a
> reboot unneccesary.  You may ask why the OS doesn't allow in-use files to be
> replaced, and the answer is that C2 doesn't allow them to, even by root
> (administrator), otherwise you could replace audit logs with doctored copies
> and cover your tracks.

Wow! is that a bogus answer. It is an outright lie. C2 has NOTHING to do with
why you can't replace executable files during operation. Who cooked that one
up? How long have you been saving that piece of crap?

The file locking mechanism in any operating system controls many access
characteristics of files. Log files are locked by logging applications. Duh!

Executable files (.DLL, .EXE, .SCR etc) are locked in Windows because they are
constantly referenced during the execution of the program. One of the ways
Microsoft tries to save page file space is to map executable segments from
within an executable file. This means that he code in the executable file need
not be in the paging file.

You answer is an outright lie. C2 has nothing to do with executables being file
locked during execution without a way to replace them on a running system. C2
has nothing to do with having to reboot after installing software. You are
lying.

 
 
 

Importance of Uptime

Post by Mike » Sat, 16 Mar 2002 14:50:31




>> What you recall and what I could find by looking at the actual bug
>> reports seem to be contradictory as well. Would you be so kind as to
>> back that claim up?
> Just download code and test it on a 2.2.x kernel if you have got one,
> I think it would work, but I am not sure after all ... it is long time
> ago.

Well I don't have one, so I can't help here.  Besides which, if it was
a problem on 2.2 as well, it would have been fixed when they fixed the
2.4 series, so current versions from 2.2 wouldn't be susceptible.

The first hit I found for it said the problem was in 2.4.1 plus an ac-
patch.  Besides -- you still haven't really justified why you're using
this as an example.

Quote:>> > 2) Don't use 1.0 versions because they are not good.
>> I never said that. The current kernel is, in fact, version 2.4.
> I understand you, but you should understand me too.

Nobody has declared that linux is perfect and does not contain any big
ol' * bugs.  The magic "1.0" is the point at which:

a) any features that are considered must-have for that version are in
   place,
b) all the new features have been tested and old features regression
   tested to find any bugs, and
b) all known bugs have been squashed.

There will still be bugs.  But as has been pointed out, they are fixed
and usually fixed very quickly.  You keep ranting about a bug that has
been squished over a year ago.  Get over it; the kernel has. ;-)

Quote:>> [stuff about running untrusted programs]
>> Many peple do tho, and since many people run windows as admin since
>> this is the default, they can cause much damage to their systems.
> The same that would run Linux as root if they used it in my personal
> opinion, and I even think that this would be even more dangerous.

This is less likely -- every linux distribution includes in the setup
a very prominent warning to NEVER USE THE ROOT ACCOUNT FOR DAY TO DAY
USAGE, or something similar.

When I installed Windows XP, it prompted me to create myself an admin
account, and then logged me into it.  It never once suggested I think
about creating another unprivileged account for day-to-day use.

*Most* people will do what they're told by the system's  installation
procedure.

Quote:> [...] And I did not meant it was a nowadays issue, just pointed an
> historical example (required by Kenneth) of Linux crashed by non root
> user. Meant nothing else.

Okay, I didn't see the start of the thread, so you can ignore my early
comment about ranting :-)  I would've picked something like the "land"
bug though, since that was a remote one-packet hard lockup ;-)

Quote:>> You provided no factual evidence and urged other people to look it up
>> for them selves in stead.
> If you want. It was not too hard, wan't it ? I told you to look for it
> if you did not believe me, just that. My own post explainned it more
> than less okay.

Providing references to reliable soures to back up your claims is just
good debating etiquette.   Forcing other people to go out of their way
to prove that you're not making things up doesn't do a lot to convince
people of your credibility.  You kept saying "it was a long time ago",
and "as far as I recall", which also doesn't strengthen your case.
--
Mike.
 
 
 

Importance of Uptime

Post by Mike » Sat, 16 Mar 2002 14:55:28



> Sincerously, for my home desktop I do not need 100%, and I am honest.
> Windows 2000 and Windows XP for my home, give me the same uptime than
> Linux, since I just have it about 10 hours a day.

Desktop systems and servers have different needs, and different users
have different needs for their desktop boxes, too.  Most only need a
few hours a day :-)

Quote:> By the way, I honestly think that rebooting a machine is good if it
> can be done without disturbing its services.

That would be an impressive feat.

Quote:> Why that uptime importance ? The Solaris and Linux servers where I
> work are booted each weekend and they still give us 100% time.

Only because they're not needed 100% of the time :-)

Quote:> I think 100% is really needed only in real time environments.

Internet servers need to be available 24 hours a day, every single day
or they risk losing traffic, which may well mean business.  Especially
true for companies that trade internationally, as rebooting the server
in the middle of the night for you  may well mean a potential customer
can't get to your site in the middle of the day, for them.
--
Mike.
 
 
 

Importance of Uptime

Post by Johan Lindquis » Sat, 16 Mar 2002 15:00:16


Fri, 15 Mar 2002 at 13:11 GMT, peering quizzically at his shoes,

Quote:>> What you recall and what I could find by looking at the actual bug
>> reports seem to be contradictory as well. Would you be so kind as
>> to back that claim up?

>    Just download code and test it on a 2.2.x kernel if you have got
> one, I think it would work, but I am not sure after all ... it is
> long time ago.

I don't really care what you think, I'm not gonna test it even tho I
do have a 2.2.20 sparc running. I might have if you had provided any
shred of evidence to this point, just to see if you were correct.

Quote:>> they call it "version 1.0".

>    So kernel 2.4.0 should have been okay after all, shouldn't it ?

"Generally" doesn't mean everyone follows the convention. I'm sure the
1.0 kernel wasn't even as stable, relatively speaking, as for example
slrn 0.9.7.3 is. The major distros didn't use the 2.4 kernel until,
what, 2.4.5? Some still default to the 2.2 kernel. That should tell
you something, don't you think?

Quote:>> you develop any software yourself at all?

>    Pointless. Anyway, as a programmer, I am not a professional, just
> hobbyst.

I just thought you had some personal standard towards this you might
like to share. Myself, I go by the 1.0 == "stable enough" convention,
but I don't do kernel programming.

Quote:>> > 2) Don't use 1.0 versions because they are not good.

>> I never said that. The current kernel is, in fact, version 2.4.

>    I understand you, but you should understand me too.

It's hard when you don't seem to make sense and try to pass off
personal opinion as facts.

Quote:>> Have you even read the kernel howto, btw?

>    Too many personal questions. This is not a test about my Linux
> knowledge, but yes, I have read it some times since VM86 patch in
> the 2.0.x series.

When you claim things like "many people run the 2.4.{0|1} kernel"
without backing it up, you'll have to expect your proficiency to be
questioned.

Quote:> I just would like that was then, in that 2.x.9 when it was really
> released, can't I have personal preferences ?

This seems to be a test of my english knowledge. I do believe I
failed.

Quote:>> No. I don't intentionally run code I don't know the purpose for.

>    That's right.

>> Many peple do tho, and since many people run windows as admin since
>> this is the default, they can cause much damage to their systems.

>    The same that would run Linux as root if they used it in my
> personal opinion, and I even think that this would be even more
> dangerous.

Windows either defaults the user to be administrator until they change
it themselves (nt4, w2k) or lets everyone be "admin" (<=w9x). Linux
does not, all the distributions aimed at the newbie users urge you to
create a normal, non-root, user account and use that instead.

Quote:>> I am the local admin on my windows box as well, hence I'm quite
>> wary of what I click on.

>    So they are a lot of persons. Running Windows is not insecure by
> definition, you just have to be careful, just a little more than
> with Linux. Anyway, NT, 2K and XP have different users as I think
> that yo do know.

The only one of those I don't know uses the admin account as the
default login is XP, and I have no personal experience of it. Of the
other, I do. With w2k you even have the option to bypass the login
screen completely if you so desire, and I don't recall there even
being a warning at that point as to the fact that it's an insecure
option.

Quote:>> You were giving an example of something that has been since fixed,
>> claiming that many people run the kernel version that it affected.

>    No, that possible many people. And I did not meant it was a

Highly unlikely, as I've pointed out.

Quote:> nowadays issue, just pointed an historical example (required by
> Kenneth) of Linux crashed by non root user. Meant nothing else.

In the context I read it, it did seem to imply that many users still
today would be affected by it. Wasn't this what you meant? If not, I
apologise for misinterpreting your article, but you really should have
just said so in your first reply to me.

Quote:>> You provided no factual evidence and urged other people to look it
>> up for them selves in stead.

>    If you want. It was not too hard, wan't it ? I told you to look
> for it if you did not believe me, just that. My own post explainned
> it more than less okay.

Yes, quite. More than less okay. I still don't believe all you say,
and while you won't back it up with factual evidence, I don't think
many others will either.

Quote:>> That is not advocacy, you are not on topic.

>    I was answering a direct question of Kenneth. And thanks for your
> help, but being or not being on topic is my own decision while this
> newsgroup is not moderated. By the way, I think I as on topic.

If you are on topic, why do you have to defend the right not to be? If
you don't care, just say so and the people here who can't stand rude
individuals can just killfile you and be done with it.

cheers,

     /Johan

--
Time flies like an arrow, fruit flies like a banana.      Perth ---> *
  1:39pm  up 13 days,  2:34,  3 users,  load average: 1.04, 1.11, 1.15
$ cat /dev/bollocks
extend extensible metrics

 
 
 

Importance of Uptime

Post by Albert Solson » Sat, 16 Mar 2002 15:01:15


Quote:> This is less likely -- every linux distribution includes in the setup
> a very prominent warning to NEVER USE THE ROOT ACCOUNT FOR DAY TO DAY
> USAGE, or something similar.
> When I installed Windows XP, it prompted me to create myself an admin
> account, and then logged me into it.  It never once suggested I think
> about creating another unprivileged account for day-to-day use.
> *Most* people will do what they're told by the system's  installation
> procedure.

   I do not agree to this one. I think that people will need root rights for
some things, and
rather than using "su -", they will logon as root. It is my personal
opinion.

Quote:> Providing references to reliable soures to back up your claims is just
> good debating etiquette.   Forcing other people to go out of their way
> to prove that you're not making things up doesn't do a lot to convince
> people of your credibility.  You kept saying "it was a long time ago",
> and "as far as I recall", which also doesn't strengthen your case.

   It is just that I am as stupid as to think that you are going to believe
what
I tell you, since I am not telling lies, it may be my fault after all.
 
 
 

Importance of Uptime

Post by Robt. Mill » Sat, 16 Mar 2002 15:05:07



> If even five minutes for a reboot is five minutes too much -- you need
> Linux.  

 I switched to Linux from OS/2 in the fall - my OS/2 system would show
signs of instability after having run about 2000 processes. Under Linux
I had just hit 3,000,000 processes when a windstorm took the power out.
Since I generated this incarnation of Linux the only times I've been
down have been due to power failures.

--

(o<     Powered by SuSE Linux
//\     Virusproof. Crashproof.  
V_/_    No MS products were used
in the creation of this message..

  8:01am  up 4 days, 17:12, 18 users,  load average: 0.00, 0.01, 0.00
processes 153019

 
 
 

Importance of Uptime

Post by Albert Solson » Sat, 16 Mar 2002 15:23:00


Quote:> Desktop systems and servers have different needs, and different users
> have different needs for their desktop boxes, too.  Most only need a
> few hours a day :-)

   Yeah :-)

Quote:> Only because they're not needed 100% of the time :-)

   Sure hehe, but the sysadms there insist on doing it even when
they can miss some Sunday 3:00 A.M. hits.

Quote:> Internet servers need to be available 24 hours a day, every single day
> or they risk losing traffic, which may well mean business.  Especially
> true for companies that trade internationally, as rebooting the server
> in the middle of the night for you  may well mean a potential customer
> can't get to your site in the middle of the day, for them.

   Some cases may be. Anyway, almost all of these big installations I
have seen have some kind of clustering system so that the individual
machines uptimes are not as critical.