File Indexing


T

tb

Using Windows 7 Professional, 64-bit.

How often does the file indexing run? I coud not find anything in the
Task Scheduler, but I am no expert of these things.

Is it possible to influence how often and at what time(s) the system
should run the indexer? How?

Thanks.
 
Ad

Advertisements

B

Bob I

From Help

The Windows search index improves the efficiency of searches by keeping
track of file names and other information for most of the files stored
on your computer. It's best to let the index run uninterrupted in the
background, but you can pause it for 15 minutes at a time.
 
Y

Yousuf Khan

Using Windows 7 Professional, 64-bit.

How often does the file indexing run? I coud not find anything in the
Task Scheduler, but I am no expert of these things.

Is it possible to influence how often and at what time(s) the system
should run the indexer? How?

Thanks.

You'll find that when the computer is new, the File Indexer is running
constantly, and your hard drive will be constantly busy. Depending on
how many files/folders you have on the drives, this will take from a few
hours to a few days to complete. Just leave your system powered up,
don't let it go to sleep, you'll find that it'll have completed its
indexing much faster. And yes, Indexing does eventually end, and it's
worth the wait.

Yousuf Khan
 
P

Paul

tb said:
Using Windows 7 Professional, 64-bit.

How often does the file indexing run? I coud not find anything in the
Task Scheduler, but I am no expert of these things.

Is it possible to influence how often and at what time(s) the system
should run the indexer? How?

Thanks.
It runs when it detects new files to index. Or, it detects
files have been deleted, in which case it might tidy up
the index a bit.

It starts from scratch, and does it all over again,
only after a long time has passed (many many days).

For example, if I restore my laptop from a backup image,
the indexer starts over again. It will process around
139000 files in three hours. Then, it stops. If I create
three files, at some point it will start running, and
process those three files. In a short burst of activity,
when you're not using the computer.

So once the majority of files are indexed, the indexer should
"settle down".

The indexer will invalidate the generated index, after some
period of time like 90 days. At least, that's how long it seems
to have gone, until one day it was at it again.

I have my indexer set, with the "back off" feature disabled.
If the indexer starts to index C: all over again, I just
walk away from the computer and let it run. With the back off
turned off, that's when it takes three hours. After the three
hours is up, I get to use the computer again. And that doesn't
happen that often.

There is a control panel for the indexer. You can leave the
control panel on display, and fool around with your files.
Like, copy a hundred files off a USB key, onto your C: drive,
then watch what happens to the indexer.

The indexer takes three hours, because it reads each file from
end to end and indexes the contents. If it was simply collecting
file names, it would be finished in 30 seconds. But it's actually
reading each file. Even if you configure the indexer to avoid
content indexing, it still does it exactly the same way. And
that's why it takes three hours.

In Task Manager, there are three processes that work together
to do the indexing. Two of them, you can raise the process
priority to "Above Normal" if you want. The third process is
a bit shy, and it gets forked afresh often enough, that you
can't really make it run faster. It's a pretty complicated
scheme, for what it's doing. When I wish to walk away, and
get the indexing done with, that's when I try stuff like
bumping up the priority. That only helps a little bit (10%).
It's still going to take three hours. I can't imagine how
painful that would be, if I had a lot of files.

I would be quite happy, if I could disable the
content indexing, but I tried that, and it didn't work.
My setting was ignored.

Paul
 
D

Dex

You'll find that when the computer is new, the File Indexer is running
constantly, and your hard drive will be constantly busy. Depending on
how many files/folders you have on the drives, this will take from a few
hours to a few days to complete. Just leave your system powered up,
don't let it go to sleep, you'll find that it'll have completed its
indexing much faster. And yes, Indexing does eventually end, and it's
worth the wait.

Not really, this is far faster than the built in search.
http://www.mythicsoft.com/page.aspx?type=filelocatorlite&page=home
 
D

Dex

It runs when it detects new files to index. Or, it detects
files have been deleted, in which case it might tidy up
the index a bit.

It starts from scratch, and does it all over again,
only after a long time has passed (many many days).
....


I would be quite happy, if I could disable the
content indexing, but I tried that, and it didn't work.
My setting was ignored.

Paul
Try removing it from 'Turn Windows features on or off' in the control panel.
 
Ad

Advertisements

P

Paul

Dex said:
On 20/09/2012 04:55, Paul wrote:

Try removing it from 'Turn Windows features on or off' in the control
panel.
What I was referring to, is the search indexing function
is supposed to have an option to index the content of
the files, or just index their file names (and not the content).
If you tell it *not* to read every stinking file from end
to end, it does it anyway.

I know how to turn off Indexing completely, so that's
not a problem. It's getting the indexer to honor user
settings that is a problem. Kinda like a car where the
steering wheel isn't firmly bolted to the steering shaft.
You can turn the wheel, but the car just goes where ever
it feels like.

And certainly, a third party search is an option.

Paul
 
D

Dex

Hmm, the link above says, "FileLocator Lite is a rebranding of Agent
Ransack for corporate environments. It has exactly the same
functionality but with a different name and logo."

In that case, FileLocator Lite will be much slower than the built in
search. Agonizingly slow, in most cases.
True. 67,000+ files and 5,400 folders taking up 300GB on a SATA II drive
in 9 seconds, bloody awful. ;)

Do it twice before a reboot and < 1 second.
 
G

Gene E. Bloch

True. 67,000+ files and 5,400 folders taking up 300GB on a SATA II drive
in 9 seconds, bloody awful. ;)

Do it twice before a reboot and < 1 second.
More like 2min 30sec on a machine with comparable statistics.

I agree with Char, except for the part about "agonizingly slow". It does
what I need and I am willing to wait...
 
Ad

Advertisements

C

Char Jackson

More like 2min 30sec on a machine with comparable statistics.

I agree with Char, except for the part about "agonizingly slow". It does
what I need and I am willing to wait...
What can I say? I'm busy! ;-)

Seriously, though, I think it's fair to compare some popular search
utilities or functions. I'll choose to compare the built-in Win 7
search, Agent Ransack (AKA FileLocater Lite), and Everything. In my
case, I'm comparing the ability to find files by name. I have no
interest in searching by contents.

What do they all have in common? You have to enter your search
parameters before they can provide meaningful results, so that part is
a wash. Once you've completed that step, however, differences start to
emerge.

Agent Ransack appears to do a single-threaded sequential search
through the designated filesystem, listing any and all results that
match the desired criteria. On my system, the volume that I search
most often is 13.3 TB and contains less than 107,000 files. While the
volume is large, the number of files is relatively small. With that
said, Agent Ransack takes about 35 seconds to search the volume.
Subsequent searches don't improve on that time, with each search after
the first also taking about 35 seconds.
PRO: I'm confident that the search results are complete and valid, and
to me that's an important consideration.
CON: The search time becomes an obstacle, especially as the number of
searches increases.

Win 7 Search is a mixed bag, for me. Indexed results appear quickly,
but I have no confidence that they are complete. Sometimes I know
something exists, and I know it exists in an indexed folder, but it
doesn't appear in the results. In those cases I switch over to Agent
Ransack. (Latest example, none of the three Win 7 systems in my office
are able to display the hosts file, though I know it exists and I know
where it's located. Rebuilding the index didn't help.)
PRO: Fast, especially for indexed locations. Good system integration,
since it's built in.
CON: Indexing system appears to be fragile, leaving me with no
confidence in the results. Results can sometimes be improved by
frequent manual rebuilding of the index, but that's time consuming and
not guaranteed to provide lasting results.

Everything isn't just fast, it's INSTANT! The time needed to search
for and display the results is zero seconds, and that holds for the
initial search as well as subsequent searches. I have confidence in
the results, but there is an edge case where the results can get out
of whack, so it's not perfect.
PRO: Fast!
CON: Networked drives require a separate search, using a web browser.

I don't do the same number of searches every day. Some days it might
only be a half dozen, while other days the number approaches 100, but
many times I'm holding the phone with one hand and have a client on
the line, so search speed is the top priority, followed very closely
by confidence in the results. For me, Everything is the clear winner,
with Win 7's Search in a rather distant second, and Agent Ransack in
3rd place only because there were only three entrants in the race.

(To keep this from being a long post, I skimmed the treetops above
when describing my impressions of the three search tools.)
 
G

Gene E. Bloch

What can I say? I'm busy! ;-)

Seriously, though, I think it's fair to compare some popular search
utilities or functions. I'll choose to compare the built-in Win 7
search, Agent Ransack (AKA FileLocater Lite), and Everything. In my
case, I'm comparing the ability to find files by name. I have no
interest in searching by contents.

What do they all have in common? You have to enter your search
parameters before they can provide meaningful results, so that part is
a wash. Once you've completed that step, however, differences start to
emerge.

Agent Ransack appears to do a single-threaded sequential search
through the designated filesystem, listing any and all results that
match the desired criteria. On my system, the volume that I search
most often is 13.3 TB and contains less than 107,000 files. While the
volume is large, the number of files is relatively small. With that
said, Agent Ransack takes about 35 seconds to search the volume.
Subsequent searches don't improve on that time, with each search after
the first also taking about 35 seconds.
PRO: I'm confident that the search results are complete and valid, and
to me that's an important consideration.
CON: The search time becomes an obstacle, especially as the number of
searches increases.

Win 7 Search is a mixed bag, for me. Indexed results appear quickly,
but I have no confidence that they are complete. Sometimes I know
something exists, and I know it exists in an indexed folder, but it
doesn't appear in the results. In those cases I switch over to Agent
Ransack. (Latest example, none of the three Win 7 systems in my office
are able to display the hosts file, though I know it exists and I know
where it's located. Rebuilding the index didn't help.)
PRO: Fast, especially for indexed locations. Good system integration,
since it's built in.
CON: Indexing system appears to be fragile, leaving me with no
confidence in the results. Results can sometimes be improved by
frequent manual rebuilding of the index, but that's time consuming and
not guaranteed to provide lasting results.

Everything isn't just fast, it's INSTANT! The time needed to search
for and display the results is zero seconds, and that holds for the
initial search as well as subsequent searches. I have confidence in
the results, but there is an edge case where the results can get out
of whack, so it's not perfect.
PRO: Fast!
CON: Networked drives require a separate search, using a web browser.

I don't do the same number of searches every day. Some days it might
only be a half dozen, while other days the number approaches 100, but
many times I'm holding the phone with one hand and have a client on
the line, so search speed is the top priority, followed very closely
by confidence in the results. For me, Everything is the clear winner,
with Win 7's Search in a rather distant second, and Agent Ransack in
3rd place only because there were only three entrants in the race.

(To keep this from being a long post, I skimmed the treetops above
when describing my impressions of the three search tools.)
To be fair, my remark about how long File Locator/Agent Ransack takes is
for the first search.

If I haven't closed FL, a second or third search (for a different name,
I mean) will go much faster, so evidently the program buffers & keeps
its data.
 
C

Char Jackson

To be fair, my remark about how long File Locator/Agent Ransack takes is
for the first search.

If I haven't closed FL, a second or third search (for a different name,
I mean) will go much faster, so evidently the program buffers & keeps
its data.
That's interesting, but I can't duplicate it here. Every search
appears to reach out and touch the filesystem in the same way. How do
I enable the caching?
 
G

Gene E. Bloch

That's interesting, but I can't duplicate it here. Every search
appears to reach out and touch the filesystem in the same way. How do
I enable the caching?
I didn't, so you shouldn't have to.

I'm going to do an experiment to make sure I didn't lie to you. I'll be
back in three minutes. Or six :)

OK, it took around 3 minutes the first time, finding 2240 *.ini files
(I'm not sure of the exact time because I was interrupted at the wrong
moment).

I closed and restarted the program so as to clear the buffer to rerun
the experiment. It took under 10 seconds for the same search, so
apparently it didn't clear its cache.

Without restarting FL, I searched for a different item (*.txt) and it
took under 10 seconds to find 3142 files.
 
C

Char Jackson

I didn't, so you shouldn't have to.

I'm going to do an experiment to make sure I didn't lie to you. I'll be
back in three minutes. Or six :)

OK, it took around 3 minutes the first time, finding 2240 *.ini files
(I'm not sure of the exact time because I was interrupted at the wrong
moment).

I closed and restarted the program so as to clear the buffer to rerun
the experiment. It took under 10 seconds for the same search, so
apparently it didn't clear its cache.

Without restarting FL, I searched for a different item (*.txt) and it
took under 10 seconds to find 3142 files.
I did a search for fred. (He wasn't lost; it's just easy to type.)
The initial search took 47 seconds. Hitting the Start button again,
with the same parameters, took 29 seconds, so I guess I'm seeing some
of what you're seeing. A 3rd and a 4th search for fred each completed
in 29 seconds, as well. Changing fred to frank, I was back at 49
seconds (close enough to 47, above). Running frank 3 more times
resulted in 29, 29, and 30 seconds.

So there is some caching somewhere, although I don't see anything in
the config that specifically explains it, nor can I match your
dramatic speed-ups. Changing the search parameters seems to clear the
cache here, while perhaps not for you. I tried unchecking the box that
allows multiple threads, but it didn't make any difference in the
times.

I'll be keeping this tool in my toolbox, but I think it's destined
never to be my first choice.
 
Ad

Advertisements

R

Rodney Pont

I closed and restarted the program so as to clear the buffer to rerun
the experiment. It took under 10 seconds for the same search, so
apparently it didn't clear its cache.
It'll be the Windows disc cache hanging onto the last data you read. As
I understand it Windows can use any spare RAM as a disc cache these
days.
 
G

Gene E. Bloch

It'll be the Windows disc cache hanging onto the last data you read. As
I understand it Windows can use any spare RAM as a disc cache these
days.
Does that cache hang on to the whole directory?

At first glance that seems unlikely, but a moment's thought makes me
realize the directory is a relatively small bunch of data.

So I'm running an experiment,

C:
cd \
dir /S > C:\Temp\HowBig.txt

which is taking about a year :)

OK,it was only a couple of minutes, and the filesize is 33.3 MB, and
about 682,000 lines, for a drive with 272 GB in use.

Not surprisingly, I have no idea how Windows and/or FileFinder store
their data, but this should give an idea.
 
Ad

Advertisements

Ad

Advertisements


Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top