Allocation of hiberfil.sys

On Mon, 24 Nov 2014 03:17:55 -0600, Char Jackson wrote:

There's little advantage in using hibernation if you are using an SSD
for the operating system and applications.

Hibernation and SSD are mutually exclusive. You're trying to lump them
together.

I like to use the term "orthogonal" in lieu of "mutually exclusive". I
co-opted the term from math & physics, where it means something like
"independent", so that a change in one quantity has no effect on the
other & vice versa, like x & y values in the plane.

In other words, it means exactly the same as what you said :-)

Anyway, for some reason I like the term, and have become a bit of a
missionary for it :-)

--
Gene E. Bloch (Stumbling Bloch)

On Sun, 23 Nov 2014 18:21:32 -0700, Jeff Barnett wrote:

I'm trying to be generous to the tech here. Maybe he was trying to
describe what really happens (AIUI): once the system restarts, the
current contents of the hibernate file are never read again, but at the
next shutdown, the then currrent contents of memory are rewritten (not
reallocated).

They are not necessarily written.

Of course I meant written when hibernation happens. Obviously it's not
going to get written unless it gets written.

I suspect that most people either always hibernate or never hibernate;
you might be an outlier, and I wrote generically, thinking you wouldn't
mind.

Actually I'm 50-50. This Windows 7 desktop computer never hibernates.
The W8 desktop always hibernates. However, that's still a very rare
occurrence, since I hardly ever use it (it's an abortive attempt at a
media center, but at least it was cheap).

As my original message says, I
hibernate maybe once a year and only in response to bad external power
events - these are desk tops. My computers normally use S3 suspend.
Unfortunately, I can't disable hibernation without losing power protection.

--
Gene E. Bloch (Stumbling Bloch)

On Mon, 24 Nov 2014 03:20:35 -0500, Yousuf Khan wrote:

The Samsung tech said that hiberfil.sys was tossed away and reallocated
whenever the system was booted. This is news to me. Does anyone here
have some more information on this?

Well, if that were the case, then hibernate would never work, if it got
tossed away whenever it booted. That's specifically the time that the
hiberfil.sys should be read.

A bit disingenuous there :-)

One would have to be on seriously mind-altering drugs to create an OS
that tossed the hiberfile before, rather than after, it was used.

--
Gene E. Bloch (Stumbling Bloch)

On Mon, 24 Nov 2014 12:04:43 -0800, Gene E. Bloch wrote:

On Mon, 24 Nov 2014 03:20:35 -0500, Yousuf Khan wrote:

The Samsung tech said that hiberfil.sys was tossed away and reallocated
whenever the system was booted. This is news to me. Does anyone here
have some more information on this?

Well, if that were the case, then hibernate would never work, if it got
tossed away whenever it booted. That's specifically the time that the
hiberfil.sys should be read.

A bit disingenuous there :-)

One would have to be on seriously mind-altering drugs to create an OS
that tossed the hiberfile before, rather than after, it was used.

Sorry, Char - I see that you covered that much earlier.

--
Gene E. Bloch (Stumbling Bloch)

On Mon, 24 Nov 2014 13:32:37 -0600, VanguardLH wrote:

Char Jackson wrote:

Hibernation and SSD are mutually exclusive. You're trying to lump them
together.

Really? Then Barnett has no problems with either since they are
mutually exclusive.

Wrong conclusion. Since they're separate questions, they have separate
answers. You're still trying to group them for some unnamed reason.

One question is whether hibernation makes sense. Each person will answer
that for him or herself.

Could you share your crystal ball or your subscription to Psychic
Friends so the rest of us know the OP understands that information?

Do you know what hibernation is? Are you aware of its pros and cons? I'm
asking because I may have given you more credit than you were due. You
really don't need a crystal ball. There's plenty of actual information
available on the topic. What questions do you have?

In my case, ...

Gee, I'm not allowed to mention related topic but you are?

You were replying to Jeff Barnett.
I'm replying to you, not to Jeff Barnett. See the difference?

Hibernation works by loading
a file into memory on Windows startup. That takes time.

It takes a whole lot less time than loading each application individually,
not to mention that loading apps individually means you also have to locate
and open the documents you previously had open.

Look at Barnett's stern reply to me. He wants only his questions
addressed. We are not to assume anything else. Barnett never said that
he required the state of his computer(s) to be the same on startup as on
shutdown. You assumed that which means that part of your reply adds
verbosity you berated me for.

Let me stop you there, since you obviously missed some key points from
Jeff's OP and his follow-ups. In case you weren't aware, a standard shutdown
closes applications and any related data files without saving them. That
usually means data loss. On the other hand, a hibernation shutdown DOES save
any open documents, (not to the general filesystem, but to the hibernation
file), preserving everything you've been working on.

Now that you have some basic knowledge of the difference, which do you
suppose is better if your goal is to not lose any unsaved work? Likewise,
which do you suppose is better if you need to account for situations where
the power may go off while no one is near the computer?

If you were able to answer both of those last two questions without moving
your lips, you're well on the road to understanding why Jeff wants
hibernation and why he doesn't care about your mythical 'long time' for
shutdown.

So, despite any concern on shutdown time in a power outage to ensure
proper shutdown within the remaining time of the UPS, there is still the
issue of writing a hiberfil.sys file. As yet, we don't know why Barnett
picked hibernation instead of a normal shutdown.

Yes, we do. It was in his OP and he repeated it several times in his
follow-ups. He cannot assure himself that he'll always be within reach of
the computer in the event of a power loss. In his case, power loss must not
be synonymous with data loss. It's as simple as that. You muddied the waters
when you didn't need to.

--

Char Jackson

Jeff Barnett wrote:

I can't disable hibernation without losing power protection.

Please elaborate on the claimed dependency. "That's news to me."

The client programs that came with the UPSes that I've used can do a
normal shutdown of Windows which includes powering off the computer.
They don't demand use of hibernation. Which UPS are you using?

Example only:
http://www.apcmedia.com/salestools/P...AGKH_R0_EN.zip
What happens during a prolonged power problem?
After the specified number of minutes have elapsed OR until the runtime
drops to the specified time to shutdown, hibernation begins. (If
hibernation is disabled, shutdown begins.)

Another example:
http://www.cyberpowersystems.com/use.../PPBE221UM.pdf
That one makes it more clear that the choice of shutdown mode is yours.

Have you been mandated by your superiors that hibernation must be
employed for a power outage and constitutes your claim that there be no
power protection without using hibernation, too?

The suggestions have been to not use hibernation with your concern about
wear levelling on the SSD. You never mentioned why you [are forced to]
use hibernation versus a normal shutdown. Do you require the state of
the computer to be restored to what it was on shutdown? Or is your
concern that the computer shuts down properly in case of a power outage
event? Yes, this is a side issue to SSD wear levelling but a related
topic as to whether hiberfil.sys should even be an issue.

On yet another side but related topic (because of your mention of UPS),
do you configure the BIOS of these workstations to start booting the
moment power has restored? Or do you leave them powered off awaiting
the return of a user for actual use? If you're concerned about SSD wear
then why have the computers running while the user is absent during
which the OS and running apps are making writes to the SSD?

On Mon, 24 Nov 2014 11:52:10 -0800, "Gene E. Bloch"
wrote:

On Mon, 24 Nov 2014 03:17:55 -0600, Char Jackson wrote:

There's little advantage in using hibernation if you are using an SSD
for the operating system and applications.

Hibernation and SSD are mutually exclusive. You're trying to lump them
together.

I like to use the term "orthogonal" in lieu of "mutually exclusive". I
co-opted the term from math & physics, where it means something like
"independent", so that a change in one quantity has no effect on the
other & vice versa, like x & y values in the plane.

In other words, it means exactly the same as what you said :-)

Anyway, for some reason I like the term, and have become a bit of a
missionary for it :-)

Agreed. Great word choice. I need to expand my vocabulary.

--

Char Jackson

On Mon, 24 Nov 2014 12:14:52 -0800, "Gene E. Bloch"
wrote:

On Mon, 24 Nov 2014 12:04:43 -0800, Gene E. Bloch wrote:

On Mon, 24 Nov 2014 03:20:35 -0500, Yousuf Khan wrote:

The Samsung tech said that hiberfil.sys was tossed away and reallocated
whenever the system was booted. This is news to me. Does anyone here
have some more information on this?

Well, if that were the case, then hibernate would never work, if it got
tossed away whenever it booted. That's specifically the time that the
hiberfil.sys should be read.

A bit disingenuous there :-)

One would have to be on seriously mind-altering drugs to create an OS
that tossed the hiberfile before, rather than after, it was used.

Sorry, Char - I see that you covered that much earlier.

No worries. As usual, your response was more succinct and on point.

--

Char Jackson

VanguardLH wrote, On 11/24/2014 1:28 PM:
Jeff Barnett wrote:

I can't disable hibernation without losing power protection.

Please elaborate on the claimed dependency. "That's news to me."

The client programs that came with the UPSes that I've used can do a
normal shutdown of Windows which includes powering off the computer.
They don't demand use of hibernation. Which UPS are you using?

I've deleted and not read the rest of your message since you are off in
a world of your own. The APC software supplied with my (and most of
their) UPS devices uses hibernation. It's not relevant whether some
other UPS does things differently. Next: if windows applications are
"open" when the computer shuts down, data can be lost. UPS software can
shut down a computer or hibernate. Now you might understand the issues.
--
Jeff Barnett

On Mon, 24 Nov 2014 12:00:09 -0800, "Gene E. Bloch"
wrote:

On Sun, 23 Nov 2014 18:21:32 -0700, Jeff Barnett wrote:

I'm trying to be generous to the tech here. Maybe he was trying to
describe what really happens (AIUI): once the system restarts, the
current contents of the hibernate file are never read again, but at the
next shutdown, the then currrent contents of memory are rewritten (not
reallocated).

They are not necessarily written.

Of course I meant written when hibernation happens. Obviously it's not
going to get written unless it gets written.

Mind blowing, there. ;-)

I suspect that most people either always hibernate or never hibernate;
you might be an outlier, and I wrote generically, thinking you wouldn't
mind.

You're probably right, so I may not be typical. My goal on this Win 8 laptop
is to do a normal shutdown as often as possible, but I frequently have
multiple documents and sites open and no time to properly close them, so I
hibernate instead. On my Win 7 desktop, however, I don't think I've
hibernated that system even once in the years that I've had it.

--

Char Jackson

Jeff Barnett wrote:

Spalls Hurgenson wrote:

Why not disable hibernation entirely and - if APC battery gets too low
- do a proper shutdown instead? Is that not an option in the APC
software? Is there a reason you need to hibernate in these situations
rather than allowing the software to perform a full shutdown when
power gets low?

In order to obey your ideas [don't use hibernate] and be properly
protected, I'd need to save all my work and shutdown each time I,
say, went to the bathroom or made a cup of coffee.

Since you have the UPS, *it* would do the shutdown *only* if needed
during a power outage. If the light switch in the bathroom also
controls the wall switch to which your computer is connected, you have
electrical wiring issues. You said the power outage occurs about once a
year. You only use the bathroom once per year? (tit for tat humor)

Is your UPS undersized to barely have enough reserve power to last the
time to detect a power outage and initiate and complete a shutdown of
Windows? I've never bought one so optimized just to handle the load for
a barely sufficient up-time. Plus batteries wane their capacity over
time so the up-time also wanes. Usually I get them so the computer and
monitor can stay up 15 minutes, or more (usually a lot more figuring for
a 5-15% degradation of the battery each year).

For example, that web form you use to enter data at some site may not
have its fields all populated again upon your return when you establish
a new connection to that site. Ever notice when you leave the page with
a web form and return to that page (Backspace) that the data you entered
was lost? With a power outage and shutdown (whatever mode), you will
lose the connection to that site. Despite using hibernation, you really
don't get the exact state you had before, especially if any app is
networked. Just because all the apps get their processes loaded into
memory on a restart when using hibernation mode doesn't mean all those
apps are going to be usable from that restored state. Oh, you say, it's
not some webmail or unimportant site that would be involved but then we
don't know if some of the app(s) you use are web-based or none are.

And no, the APC software offers no such options.

The UPSes I've had do not force me to use hibernation. In my other
reply, I cited the manual for Powerchute (APC's client software) that
mentions hibernation or normal shutdown can be used. It doesn't make
clear where to configure the client to select which shutdown mode but
they imply there is a choice.

http://msdn.microsoft.com/en-us/libr...(v=vs.85).aspx

Didn't see anything there about specifying hibernation for the shutdown
mode.

http://msdn.microsoft.com/en-us/libr... =vs.110).aspx

That's a .Net method (programming library) so there must be an
underlying system API that can accept parameters as to how it should
shutdown. So it looks like the client can specify hibernate or
shutdown. The example CyberPower manual mentions a setting to choose.
You'll have to dig around the APC manual or the client's config screens
to see where they have the setting to which they refer.

It sounds like you believe that the 50GB hiberfil.sys will occupy the
same blocks on the SSD all the time. Your worry is that those 50GB of
blocks will remain in near-pristine condition while the remaining
blocks of the SSD endure all the usual read-writes and - because they
are pulling "double-duty" for the never-changing 50GB of blocks, will
wear out faster.

This is exactly what I fear. The hiberfil.sys is rarely if ever used on
my desktops - it's like the premium on an insurance policy.

Cluster allocation (a object) doesn't change in the file table (a
logical structure) is independent of physical allocation on the storage
media. Review wear levelling for SSD media, like:

http://en.wikipedia.org/wiki/Wear_leveling
http://searchsolidstatestorage.techt.../wear-leveling
http://thessdguy.com/how-controllers...wear-leveling/
http://www.networkcomputing.com/stor...a/d-id/1097528

Wear levelling is done in the hardware device, not in the OS or how it
manages logical assignment of clusters to files. Wear levelling means
every block on your SSD will eventually get used (so it eventually
incurs the double write operation: zero data out of the old block, write
new data to a different block).

Aside (and warning to Jeff of a related but side topic): Because cells
go bad is why there is remapping of them to new cells. Remapping slows
the device so the more there are then the slower the device. Eventually
the SSD catastrophically fails when there is no more reserve remapping
space available. Backups (to other media) are important.

That the hiberfil.sys file occupies the same clusters within the MFT is
independent of which physical blocks of memory in the SSD that the file
currently occupies, or what blocks in the SSD the file will occupy in
the next write and so on. Static in the MFT doesn't not mean static in
the SSD.

On Mon, 24 Nov 2014 14:28:17 -0600, VanguardLH wrote:

[Deleted 5 paragraphs of fluff. You're very verbose!]

On yet another side but related topic (because of your mention of UPS),
do you configure the BIOS of these workstations to start booting the
moment power has restored? Or do you leave them powered off awaiting
the return of a user for actual use? If you're concerned about SSD wear
then why have the computers running while the user is absent during
which the OS and running apps are making writes to the SSD?

Another red herring. Surely you make a distinction between SSD writes by an
idle system versus SSD writes from a system being actively used, at least in
general? You seem hell bent on keeping the water as muddy as possible.

--

Char Jackson

Jeff Barnett wrote:

VanguardLH wrote, On 11/24/2014 1:28 PM:
Jeff Barnett wrote:

I can't disable hibernation without losing power protection.

Please elaborate on the claimed dependency. "That's news to me."

The client programs that came with the UPSes that I've used can do a
normal shutdown of Windows which includes powering off the computer.
They don't demand use of hibernation. Which UPS are you using?

I've deleted and not read the rest of your message since you are off in
a world of your own. The APC software supplied with my (and most of
their) UPS devices uses hibernation. It's not relevant whether some
other UPS does things differently. Next: if windows applications are
"open" when the computer shuts down, data can be lost. UPS software can
shut down a computer or hibernate. Now you might understand the issues.

Then APC must be lying in their Parachute manual (what you conveniently
snipped out since it would detract from your lambaste).

Char Jackson wrote:

On Mon, 24 Nov 2014 10:07:09 -0500, Spalls Hurgenson
wrote:

Why not disable hibernation entirely and - if APC battery gets too low
- do a proper shutdown instead? Is that not an option in the APC
software? Is there a reason you need to hibernate in these situations
rather than allowing the software to perform a full shutdown when
power gets low?

I think the concern is that that would result in data loss, whereas
hibernation preserves in-use data.

For networked or web-based apps, too? Don't think so. Your host will
lose the connection so it's unlikely on reboot that the web form with
all your entered data will remain usable. Some web apps require a
persistent connection.

On Mon, 24 Nov 2014 15:27:13 -0600, VanguardLH wrote:

Jeff Barnett wrote:

Spalls Hurgenson wrote:

Why not disable hibernation entirely and - if APC battery gets too low
- do a proper shutdown instead? Is that not an option in the APC
software? Is there a reason you need to hibernate in these situations
rather than allowing the software to perform a full shutdown when
power gets low?

In order to obey your ideas [don't use hibernate] and be properly
protected, I'd need to save all my work and shutdown each time I,
say, went to the bathroom or made a cup of coffee.

Since you have the UPS, *it* would do the shutdown *only* if needed
during a power outage.

If you configure your UPS to do a standard shutdown, does it know how and
where to save all opened documents?

[Snipped multiple paragraphs of fluff.]

--

Char Jackson