Log Off VS Restart

[xrees]

A simple question. Varied answers I'm sure.
A client recently asked

Should we "Log Off" or "Restart" at the end of the day.

Which is better?
And why?

Log Off
1. Gets you out of all the programs you were working in, shutting down
any data access to email, documents etc. so that backups can perform
without running into "open files".
2. Shuts down any applications that were run on LOGIN.

Restart
1. Same as Log Off PLUS
2. Clears the Page File?
3. Clears RAM

What about electrical surges? Does a restart actually cut power
temporarily to hard drives, MB, etc. ? Ie is there more wear and tear
on a workstation performing daily RESTARTS as opposed to daily LOGINS
and LOGOFFS?

TIA
Luke

 

[JS]

Depends on what the janitor walking around in the middle of the night can
see.
If the PC are connected to an UPS and a good firewall, then I guess you
could Log Off and leave
the systems on. No UPS, Shut it down (Power Off).

JS

 

[Wiley Coyote - N2K]

A few things. First, why would one want to RESTART at the end of the day? I
am going to assume the question would be "Should I log off or SHUTDOWN at
the end of the day". Just guessing.

In any case, logging off (in it's simplicity) closes network connections,
updates the registry hive HKCU, closes open programs etc. In addition, NTLSA
(NT Local Security Authenticator) causes the Kernel to wait for the
CTRL-ALT-DEL combination to allow someone to Log on and be authenticated. In
addition to starting services and system apps.

Shutting down (or restarting) essentially does the same thing, in addition
to starting an entire new Windows Environment. However, certain events will
not occur, such as starting user sessions for things like Anti-viral (spam
etc) applicaitons, Office type startup apps (such as Fast Finders etc).

When a user is logged off, the system is still "Seeable" on the network. As
such, a would be hacker could use tools to probe the machine (such as
netstat, nbtstat etc). These type tools can reveal much about the system for
example, the Domain/Workgroup, what ports are open on the system, user
credentials etc. Having that said, when a user is logged on, the same
information can be probed anyway.

I think one of the answers would be "What is the companies security
requirements?". If security is (and it should be) an issue, then Shutting
Down would be the better option. This would be (from a security concern) the
same as "Unplugging from the network". Albiet this is a lame and simple
analogy, but I think the point is made.

I realize that this is rather vague, but there are many other types of
issues that could be dicussed and then we would have to start writing
novells.

Cheers,

Wiley.

 

[DanS]

A few things. First, why would one want to RESTART at the end of the
day? I am going to assume the question would be "Should I log off or
SHUTDOWN at the end of the day". Just guessing.

I'm quite sure the OP did not mean shutdown....

This domain looks to be setup to do user backups after hours, so shutdown
is not an option.

 

[DanS]

A simple question. Varied answers I'm sure.
A client recently asked

Should we "Log Off" or "Restart" at the end of the day.

Which is better?
And why?

Log Off
1. Gets you out of all the programs you were working in, shutting down
any data access to email, documents etc. so that backups can perform
without running into "open files".
2. Shuts down any applications that were run on LOGIN.

Restart
1. Same as Log Off PLUS
2. Clears the Page File?
3. Clears RAM

What about electrical surges? Does a restart actually cut power
temporarily to hard drives, MB, etc. ? Ie is there more wear and tear
on a workstation performing daily RESTARTS as opposed to daily LOGINS
and LOGOFFS?

TIA
Luke

I know you really did mean restart Luke.

The question is, have you experienced any difference if someone chooses
one or the other ? Has there been any issue's with locked files when just
logging off. Since network shares are still active even though noone's
logged on that particualr PC, other users *COULD* be accessing the files
on a remote drive, which MAY cause a problem.

You can always drum up a discussion of which is better, on 24/7 or on
when only using..... should I logoff or just lock the PC, etc.

Unfortunately I can't answer the question about cutting power. I haven't
looked at PC architecture in depth. I would venture a guess that there is
less of a 'surge' on just a reset. How much ? Don't know. Only because
the HD may still be spinning since the reset is of very short duration.
There will not be the initial draw needed to start it spinning. Also,
power supply capacitors will still be charged to be able to supply *some*
of that initial oomph needed.

Anyway you look at it, whatever any differences may be bwteen the two,
they'd probably wash each other out anyway.

It ends up being personal preference.

 

[xrees]

I know you really did mean restart Luke.

The question is, have you experienced any difference if someone chooses
one or the other ? Has there been any issue's with locked files when just
logging off. Since network shares are still active even though noone's
logged on that particualr PC, other users *COULD* be accessing the files
on a remote drive, which MAY cause a problem.

You can always drum up a discussion of which is better, on 24/7 or on
when only using..... should I logoff or just lock the PC, etc.

Unfortunately I can't answer the question about cutting power. I haven't
looked at PC architecture in depth. I would venture a guess that there is
less of a 'surge' on just a reset. How much ? Don't know. Only because
the HD may still be spinning since the reset is of very short duration.
There will not be the initial draw needed to start it spinning. Also,
power supply capacitors will still be charged to be able to supply *some*
of that initial oomph needed.

Anyway you look at it, whatever any differences may be bwteen the two,
they'd probably wash each other out anyway.

It ends up being personal preference.

Thanks to everyone who replied.

The dialogue with the client had started because of the old question
"should we shut our pcs down at night?" I used to be of the opinion
that yes, to save on electricity, keep the amount of heat down etc. it
would probably be best to do so. Now, after years of seeing systems
lose their drives "way before their time" usually on startup in the
mornings, I'm not so sure.

So i offered my opinion, that it would probably be better if they left
their pcs on for the duration of the week, logging off so as to make
sure that everyone was out of all possible data files, dbs, etc. Then
to shut them off on the weekends. So when the question came to me of
whether a restart would do the same thing and which was better, I was
sort of caught off guard. Never thought of it really. Was wondering
if anyone else ever had.

Which lead me to this group and the questions about loss of power, etc.
I'm inclined to believe that you're probably right, that the amount of
time that the drives go without power is miniscule and hence very
little eletrical stress is imposed on all the components. So that a
restart or logoff (other than all those other things that occur on
startup already mentioned) is really all a matter of choice and not
benefits.

Thanks again
Luke

 

[w_tom]

Does power cycling harm electronics? Yes. And then we add numbers.
For example, a disk drive with an extremely low power cycle number was
40,000 power cycles. 40,000 power cycles five times every day to
that disk drive is - 22 years. Yes, power cycling is destructive if
your computer is for over 20 years of operation.

Power switches are typically rated for 100,000 cycles. 100,000 power
cycles - seven times every day - is 39 years. So yes, power cycling is
destructive - and who really cares?

It is a classic example. Many know - and then forget to put that
knowledge to the reality of the numbers.

So what is a digital transistor doing? Power cycling repeatedly. Once
transistors were rated for number of switching cycles. No longer
because the numbers are so large as to not be relevant. Take a look at
numbers for original Pentium microprocessor. It must go from less than
1 amp current to tens of amps current. Even worse, it must do this in
microseconds. A massive power surge to both microprocessor and
adjacent power supply - that occurs in normal operation. Power off and
these stresses numbers are reduced.

But again, once we add the missing perspective - the numbers - then
power cycling is a problem with too many hours of operation. IOW turn
it off or hibernate it when done. If power cycling was so destructive,
then other electronics (TVs, radios, VCRs, etc) would also always be
left on. Include the numbers; power it off when done.

 

[xrees]

I appreciate the post w_Tom.

I remember someone telling me something funny about numbers one time.

There's lies, damn lies and then there's statistics.

It doesn't really matter if hard drives were rated for 1,000,000 power
cycles. What manufacturers can recreate in the lab and what actually
happens in the real world are two different things. Otherwise you
wouldn't have RMA departments at Maxtor, Fujitsu, etc.

Theory is wonderful. Especially when tempered with experience. I
would say the bulk of dead hard drives that I've come across have
happened on startup in the morning. They were working fine, the
evening before and nada the morning after. Stressed from too much time
away from the flow of electrons? Or zapped by a massive surge of
electricity? I don't know. But what I do know, is that for the bulk
of the systems that we look after, the ones that don't power down at
night seem to have less troubles than the ones that power cycle their
machines every night.

Again. Just my personal experience.

Oh, and I'll have to concur with you on personal electronics. Tho
there aren't too many spinning parts in my TV. And my toaster doesn't
have to crisp my toast in a nanosecond either.



Regards
Luke

 

[w_tom]

"Lies, damn lies, and statistics" is the preception of MBA school
graduates who somehow are experts and yet never learned even basic
science. Yes, numbers can lie when the human does not first learn
basic concepts - and does not temper that knowledge with experience.
Therefore believe who we 'feel' is telling the truth? Nonsense.
Without numbers, then lying becomes many times easier. Numbers may not
be perfect. But no numbers is a symptom of more and outright
deception. Numbers are a commitment that may even cause the
manufacturer to change those numbers as more experience is obtained.

That 40,000 number was from a disk drive manufacturer who once
claimed 100,000 cycles. As I said, 40,000 was one of the lowest
numbers I ever saw for a disk drive because manufacturer (in response
to a free market and the need to be responsible) lowered that spec to
40,000. Would we know this? Not if numbers were withheld. Anybody
can 'lie or decieve' routinely if numbers are not provided. Written
numbers are what hold the manufacturer to honesty.

In another computer component - power supplies - some manufacturers
never provide numbers because their power supplies are being marketed
to computer assemblers - who would not know what those numbers mean
anyway. Those 'dumped into the market' (and therefore defective)
power supplies are missing essential functions. What is the first
indication? They provide no numbers. What is a second indication?
Strange failures that the computer assembler then blames on things he
does not understand such as surges, room temperature, improper power
removal, too few fans, or power cycling.

Do numbers always mean every number is always accurate? Of course
not. But when no numbers are provided, then a product more often is
even worse. Look at those more highly regarded power supply
manufacturers. Responsible manufacturers provide essential internal
functions, and also (or therefore) provide the numbers. No spec
numbers for a power supply often means supply is being marketed to
'computer experts' who don't even know how electricity works.
Those are the same people who would then hype sound bytes: "lies,
damn lies, and statistics".

Finally, do you disassemble or repair to learn why electronics fail?
Power cycling is not a typical reason for failure. However many
defects (hours of operation, manufacturing defects, etc) don't become
apparent until a next power cycle. For example, defects created by
excessive hours of operation can finally cause component failure do to
a so gentle and what is normally irrelevant shock from power up. The
symptom is apparent often during power up. Therefore that is the time
most failures are observed. Those who don't first learn why failures
occur then *assume* power up caused the problem.

Why do we leave parts of a satellite completely powered off? So that
that duplicate system will not fail. Power down again a way to
perserve the backup system. Leaving powered would only degrade that
backup system.

But again, answer demands things such as numbers and asking questions
containing the word why. Power up rarely causes failure. More often
failure is due to manufacturing defects, static electric discharges,
abuse, etc. These failures then become apparent during a later power
ups.

Power cycling is what the user sees even when the failure was created
during normal operaton - and yet is only apparent during next power
cycle.

I have yet to find a machine damaged by power cycling. I even fix
power supplies only to learn why supply failed. One supply would not
start because a pullup resistor failed during normal operation.
Without 'first asking why', then one would assume power cycling
cause that partiular failure. Again, responsible manufacturers provide
numbers such as 40,000 and 100,000 because those are numbers he is
committing to in writing. Power cycling six times every day for 29
years - who cares? But yes, power cycling does cause failures. An
answer that means nothing without a grasp of above simple concepts that
an never be explained in 'sound byte' reasoning.. An answer made
irrelevant once we provide numbers. No numbers is a first symptom of
defective products from a scam manufacturer. Those numbers are more
than theory. Numbers are a manufacturer literally committing to
reality.

 

[Pez D Spencer]

"Lies, damn lies, and statistics" is the preception of MBA school
graduates who somehow are experts and yet never learned even basic
science. Yes, numbers can lie when the human does not first learn
basic concepts - and does not temper that knowledge with experience.
Therefore believe who we 'feel' is telling the truth? Nonsense.
Without numbers, then lying becomes many times easier. Numbers may not
be perfect. But no numbers is a symptom of more and outright
deception. Numbers are a commitment that may even cause the
manufacturer to change those numbers as more experience is obtained.

*********snipped for re-post brevity**********

I have yet to find a machine damaged by power cycling. I even fix
power supplies only to learn why supply failed. One supply would not
start because a pullup resistor failed during normal operation.
Without 'first asking why', then one would assume power cycling
cause that partiular failure. Again, responsible manufacturers provide
numbers such as 40,000 and 100,000 because those are numbers he is
committing to in writing. Power cycling six times every day for 29
years - who cares? But yes, power cycling does cause failures. An
answer that means nothing without a grasp of above simple concepts that
an never be explained in 'sound byte' reasoning.. An answer made
irrelevant once we provide numbers. No numbers is a first symptom of
defective products from a scam manufacturer. Those numbers are more
than theory. Numbers are a manufacturer literally committing to
reality.

fascinating analysis.

yes, i've always found that you have to single out the actual source of
the failure before you can deal with the actual problem effectively.
if you don't, then you may replace a whole computer when it was just
the heat sink that failed.

if you don't know which character is hurling that devastating meteoric
shower at your character, then you can't respond properly. at least,
it makes sense in rpgs...

 

[xrees]

-sigh-

The biggest surge of current a typical system will see is on startup.
Not when its operating.
At startup.
There is something called an "inrush current".
But then again you know all about that, don't you Tom.

http://groups.google.ca/group/alt.c...+surges+startup&rnum=1&hl=en#1a108521d0330fb6

Coupled with the fact that everything needs to be enegized on the
board, you have actual mechanical items (fans, cdroms and finally the
hard drive platters) that have to overcome inertia to start rotating
once again. All these things, in all the infinte permutations possible
lead to the single most vulnerable time for a computer to fail. If it
weren't so, all the servers in the world would be powering down to
"save on electricity". It just ain't so.

So we can agree to disagree. You keep power cycling your power
supplies and disecting them to find out it was that pull up transistor
that really died during normal operation but you only found out about
it after power cycling, while I keep mine humming and happy.

Thanks for the reply
Luke

 

[Pez D Spencer]

-sigh-

The biggest surge of current a typical system will see is on startup.
Not when its operating.
At startup.
There is something called an "inrush current".
Thanks for the reply
Luke

from my experience, after thinking long and hard about this question
and the responses, computers that fail on startup fail not because of a
surge of electricity, but because of accumulated dust that burns up and
shorts everything on powerup.

which leads me to the question: how often do you guys open the case
and clean the internals? what's the best way to do it? i'm afraid
that using a compressed air aerosol can will just dislodge the dust and
put it somewhere worse.

 

[w_tom]

The biggest surge of current a typical system will see is on startup.
Not when its operating.
At startup.
There is something called an "inrush current".
But then again you know all about that, don't you Tom.

sigh ....
Even 1950s televisions contained a "Inrush Current Limiter". Limits
current during startup - as others did not first learn before
automatically assuming power up is destructive. Been doing this stuff
for too many decades. There is no massive current inrush as so often
proclaimed by those who do not even know of the current inrush limiter.

Again, power up is not this big massive and destructive event as
promoted by those who don't even know of the current inrush limiter.
Maybe first do some research? You are the one who said, "Theory is
wonderful. Especially when tempered with experience." Then why did
you not know of inrush current limiters - a technology so old as to be
seen in 1950s TVs?

Yes power up is destructive - especially when assumptions are made
without numbers and experience. Then we apply those numbers and a few
decades of experience. That destructive event now becomes totally
irrelevant. How many disk drives over many decades did you work with
at the IC level? Just like a TV, stereo, radio, etc - turn it off when
done so that hours of operation do not create weaknesses and failures.
Failures that are mostly noticed during power up. Failures that were
created during normal operation. But then you know all about 'inrush
current limiters' before posting?

 

[w_tom]

from my experience, after thinking long and hard about this question
and the responses, computers that fail on startup fail not because of a
surge of electricity, but because of accumulated dust that burns up and
shorts everything on powerup.

which leads me to the question: how often do you guys open the case
and clean the internals?

A computer that cannot operate normally in a 100 degree F room with
acceptable numbers of dust balls generally has other problems. Heat is
often a diagnostic tool to find intermittent, marginal, and slow
failing hardware. Yes, the numbers say that hardware degrades faster
in warmer environments. And then we again put those numbers into
perspective - temper the theory with both numbers and experience.
Higher heat in a 100 degree room or due to dust balls is not
destructive.

Dust becomes a problem when, for example, all airflow between
heatsink fins is obstructed. Then a toothpick or equivalent maybe
necessary to clear the obstruction. Also using too many fans inside a
PC will cause excessive dust buildup. Most all systems are
sufficiently cooled with only one chassis fan. The second fan may
result in only single digit temperature reductions. Again,
perspective. The single digit improvement is not significant.

Long before temperature causes hardware failure, it instead causes a
temporary hardware crash. Heat is not destructive. Heat is a tool to
find weak or failing hardware often before that permanent failure
occurs. When systems get too hot, they just perform erratically -
typically no permanent damage. Others then try to cure symptoms with
more fans. As the old engineers once taught me about semiconductors -
it's only too hot when you leave skin. Electronics must be designed to
also work just fine (not erratially) inside a 100 degree F room.

 

[Pez D Spencer]

A computer that cannot operate normally in a 100 degree F room with
acceptable numbers of dust balls generally has other problems.

just for clarification, i meant that i've experienced power supplies
burning out because of an extreme buildup of dust/particulate that ends
up burning up within the power supply. at least, that's what i think
happened with my ex's computer from all the dust and kitty-litter dust
that was in the same room with the computer. (she didn't take care of
it at all.) and, from my friend's estimation it also happened to him
after years of running it without cleaning out the internals.

i only have the one main fan from the power supply and the heat sink
fan which has an open intake on the case in line with the heatsink.

would someone normally, during years of normal operation, have to open
up the case and clean out from around the heatsink and the mobo?

thx.

 

[w_tom]

Dust is generally not electrically conductive; would not short out a
power supply. As long as power supply fan is moving air, dust should
not obstruct airflow sufficiently to cause burning.

But then if kitty litter was inside, one could only wonder what else
had wandered in. Once saw a picture of a snake inside the PSU.

Long before dust had become so dense as to cause power supply
failure, the fan would have failed. No fan (no PSU airflow) sometimes
means power supply component failure - only because with a fan, a PSU
manufacturer then used undersized (smaller) heatsinks.

The only reason I clean out dust is because I was inside doing
something else, don't want my hands dirty, and don't want to breath
that dust.

Dust that is metallic - now that dust would concern me.

 

[xrees]

groan...

I mentioned "inrush current" as another possible reason that failures
may happen at start up. A great discussion about Inrush Current
Limiters can be found here:

http://www.ametherm.com/Inrush_Current/inrush_current_faq.html

So what I understand from the above page is that these devices limit
the "inrush current". But that they have to cool down after initial
power up. Typically about a minute. News to me. Tho I have heard of
rapid power cycling being a no-no.

So what happens if you rapidly "power-cycle" a power supply? Sounds
like a massive "inrush current" to me. Computer manufacturers might be
wise to actually state that somewhere in the product manuals. I've
been doing this for quite some time and I honestly didn't know this.

Disk drives at an IC level? Very few. Mostly in some of my
troubleshooting courses in the mid 80's. Why do you ask?



Luke

 

[w_tom]

You have learned there is no massive inrush current due to an unknown
part that was standard even in 1950s. Step by step, learning reality
about power cycling myths - by first learning about features that were
defacto standard in power supplies long ago. Now you ask "what happens
if you rapidly "power-cycle" a power supply?" Well, obviously
electrolytic capacitors are not fully discharged, so again, no massive
inrush current. But again, that is part of a power supply design that
makes inrush current surges irrelevant.

Furthermore, another protective circuit in power supply controller
locks out the power supply. Just another protection layer on that makes
'destructive power cycling' mythical. Just another fact that is never
learned by those who declare by speculation that power cycling is
destructive. Some would 'feel' rather than first learn basic electrical
concepts. But again, where are underlying facts to prove power cycling
so destructive? A myth purveyor forgets to learn about things such as
an inrush current limiter - a technology that was standard even in the
1950. A technology that old and yet still those 'computer experts'
don't even know what it is? No wonder they promote this wild
speculation about power cycling.

Those who preach 'damage from power cycling' forget to first learn
the 'whys'. A previous example demonstrated why a power supply would
not start: due to damage created by normal operation. The naive would
immediatly proclaim that failure was from power cycling - because he
did not learn about damage created by normal operation. Having not
learned both theory and experience, some don't even know about 1950
technology - the inrush current limiter. They assume capacitors
immediately discharge. They don't know about other protective circuits
that also make power cycling irrelevant.

Many clone computer assemblers don't know how electricity works;
don't even know about inrush current limiters. They even use
shotgunning. They purchase supplies only on watts and dollars. Then
when these power supplies fail due to missing essential components,
these computer assemblers blindly blame power cycling or speculate
about surges. IOW they assume and then proclaim that assumption as
fact. In reality, failure was due to a technically naive human. He
did not first learn underlying technologies; used only dollars to make
a design decision. This technical ignorance creates myths about
'destructive power cycling'. So called 'computer experts' who did not
even know about an inrush current limiter somehow knows inrush surges
must exist.

Power cycling is destructive. When we apply details such as 'the
numbers', then power cycling is irrelevant. Power cycling is only
promoted as a problem when assumptions without numbers are represented
as facts.

 

[xrees]

uh...that's not what the faq says Tom:

"Despite the economic and design simplicity advantages Surge Limiters
offer for inrush current suppression, there are some situations when an
active circuit may provide a more appropriate solution. For example,
active circuits can sometimes be a better choice than Surge Limiters
when "hot restart" capabilities are critical. The advantage is related
to the cool down/recovery time required for the resistance of the Surge
Limiter to increase sufficiently to provide the required level of
inrush current protection. If the power drops out for a period shorter
than the specified recovery time, say, for a few seconds, the
thermistor will not have an opportunity to cool down and return to its
initial level of resistance. When the power comes back on, the
resistance is too low to provide sufficient inrush protection, so
circuits could be damaged, fuses blown, etc."

Sounds like "inrush current" can get by your "made in the 50's" device.
And when it does, I would imagine it "takes no prisoners".

Luke

 

[w_tom]

If inrush current limiters were ineffective, and since surges so
easily cause electronics damage, and since we have not even discussed
other technologies that make such surges irrelevant; better leave all
TVs and radios powered. Power on surges were also destroying all
electronic even in the 1950s when tubes had filaments routinely burned
out by power surges. Oh! Power on was not destroying them either?

Provided earlier were the numbers. Numbers confirmed by real world
eperience AND by tracing the source of failures. Numbers said to be
wrong only by using lawyer logic - also called spin. The reader has
been presented layer after layer of reasons why destructive power
cycling is only myth. Once manufacturering numbers, technical
knowledge, and experience is applied, then damage due to power cycling
is totally irrelevant.

If power cycling was so destructive, then never turn off lights,
radios, or TVs. Ironic that one who did not even know about inrush
current limters is now using lawyer logic to claim inrush current
limiters don't work. Well ICLs worked just fine in the 1950s. And
that only for supplemental protection. Other circuits did not suffer
from power cycling even without ICLs. Since then, electronics got even
better. IOW power cycling was not a problem in the 1950s. It was and
is never a problem with radios, TVs, or light bulbs. Why suddenly does
lawyer logic prove all that wrong?

It's called a myth. Promoted using Rush Limbaugh type propaganda.
Power electronics off when done. Ignore power cycling myths; promoted
without technical grasp or numbers. Myths promoted without even
knowing a 1950 technology called inrush current limiter.