Why does my HD access light blink steadily?

  • Thread starter Thread starter mailbox
  • Start date Start date
M

mailbox

Even when my new computer is idle,
-- as far as I know -- the hard
drive access LED flashes briefly
about every two seconds under Linux,
and about once per second when I'm
running Windows XP. I assembled it from
components I puchased online. It
has an MSI motherboard PM8M-V.
My wife's computer, also custom
assembled a few years ago, doesn't
do this, nor does the Dell I use
at the office. What gives?
 
Even when my new computer is idle,
-- as far as I know -- the hard
drive access LED flashes briefly
about every two seconds under Linux,
and about once per second when I'm
running Windows XP. I assembled it from
components I puchased online. It
has an MSI motherboard PM8M-V.
My wife's computer, also custom
assembled a few years ago, doesn't
do this, nor does the Dell I use
at the office. What gives?
Click to expand...

Background activity such as flushing buffers and (in the case of XP, the
fast search facility).

I'm surprised to hear of XP machines that don't exhibit this behaviour. You
don't get it on Win9x.

Dave
 
Even when my new computer is idle,
-- as far as I know -- the hard
drive access LED flashes briefly
about every two seconds under Linux,
and about once per second when I'm
running Windows XP. I assembled it from
components I puchased online. It
has an MSI motherboard PM8M-V.
My wife's computer, also custom
assembled a few years ago, doesn't
do this, nor does the Dell I use
at the office. What gives?
Click to expand...
I've never had an XP setup that flashed consistently unless it had a
virus. Normal XP behavior I've seen may have what you are seeing for
periods of 5 or 6 seconds as part of a random flashing due to XP
background, but then always reverts to a different pattern, also
containing periods of 10 or more seconds with no activity.
IOW, no detectable pattern.
Of course the software I run probably is entirely different than
yours. I'm just telling you my experience.

--Vic
 
Vic said:
I've never had an XP setup that flashed consistently unless it had a
virus. Normal XP behavior I've seen may have what you are seeing for
periods of 5 or 6 seconds as part of a random flashing due to XP
background, but then always reverts to a different pattern, also
containing periods of 10 or more seconds with no activity.
IOW, no detectable pattern.
Of course the software I run probably is entirely different than
yours. I'm just telling you my experience.
Click to expand...


The fact that I get regular, though slightly different flash rates from
the HD access light under both Windows and Linux suggests that it's
not a virus. It's hard to imagine a virus that would infect two
different operating systems identically. This is the key thing about
this phenomenon: that it occurs under both Windows and Linux. That's
why I figured that it involved low-level hardware or firmware at the
disk controller level.
 
The fact that I get regular, though slightly different flash rates from
the HD access light under both Windows and Linux suggests that it's
not a virus. It's hard to imagine a virus that would infect two
different operating systems identically. This is the key thing about
this phenomenon: that it occurs under both Windows and Linux. That's
why I figured that it involved low-level hardware or firmware at the
disk controller level.
Click to expand...

In both cases, it could be a background file-indexing task running. In Windows
it is sometimes called Find Fast and can it be turned on & off from Control
Panel. I'm not in linux at the moment but "man locate" may shed some light on it
there.

HTH
Bob
 
Robert said:
In both cases, it could be a background file-indexing task running. In Windows
it is sometimes called Find Fast and can it be turned on & off from Control
Panel. I'm not in linux at the moment but "man locate" may shed some light on it
there.
Click to expand...

Well, the "updatedb" command that is used to update the 'locate"
database is typically run once a day. I've already done a "crontab -l"
at the root level and found no scheduled jobs. I just did a newsgroup
search of linux.redhat and it turns up a query in 2003 about the
blinking hard drive light. No indication how it was resolved, though.
So maybe it is an operating system issue. I guess I'll have to figure
out how to use "ps" to capture _very_ transient processes.
 
Well, the "updatedb" command that is used to update the 'locate"
database is typically run once a day. I've already done a "crontab -l"
at the root level and found no scheduled jobs. I just did a newsgroup
search of linux.redhat and it turns up a query in 2003 about the
blinking hard drive light. No indication how it was resolved, though.
So maybe it is an operating system issue. I guess I'll have to figure
out how to use "ps" to capture _very_ transient processes.
Click to expand...

Another possibility could be any process that performs a logging function. I
don't have any names to suggest at the moment as it's been a long time since I
installed and/or made any setup mods to my own os'es, but you could search
C:\Documents and Settings (incl subfolders) for any *.log file that has been
modified on the current date. That might offer a clue. Also unplug your ethernet
connection in case random input from the Internet is causing this.

Bob
 
.... snip about why HD access occurs when idle ...
Also unplug your ethernet connection in case random input from the
Internet is causing this.
Click to expand...

That is one advantage of dial-up access. I download the mail and
usenet, access a few selected web pages, go offline and
disconnect. Now no vandals can reach me.

--
Some informative links:
<<http://www.geocities.com/nnqweb/>
<http://www.catb.org/~esr/faqs/smart-questions.html>
<http://www.caliburn.nl/topposting.html>
<http://www.netmeister.org/news/learn2quote.html>
<http://cfaj.freeshell.org/google/>
 
CBFalconer said:
... snip about why HD access occurs when idle ...
Click to expand...

In case there is any confusion about it, I had suggested that only for testing
purposes. He should naturally plug it back in after checking to see what
happened, if anything, to the HD activity.
That is one advantage of dial-up access. I download the mail and
usenet, access a few selected web pages, go offline and
disconnect.
Click to expand...

Sounds like a big PITA to me. said:
Now no vandals can reach me.
Click to expand...

They could be hiding in your backyard at this very moment. Better make sure that
door is locked!<VBG>

Bob
 
Well, the "updatedb" command that is used to update the 'locate"
database is typically run once a day. I've already done a "crontab -l"
at the root level and found no scheduled jobs. I just did a newsgroup
search of linux.redhat and it turns up a query in 2003 about the
blinking hard drive light. No indication how it was resolved, though.
So maybe it is an operating system issue. I guess I'll have to figure
out how to use "ps" to capture _very_ transient processes.
Click to expand...

If you have a journaling file system like EXT3 it has to save the journal. I
think the default is 5 seconds for ext3. You should have a journalling
filesystem if the release or update you use is less than five years old.

--
A certain thing in this world is if you say Jews are inconsequential then
Jews will start making claims of Jewish power they would call antisemitic if
a non-Jew had said them.
-- The Iron Webmaster, 3713
nizkor http://www.giwersworld.org/nizkook/nizkook.phtml
Old Testament http://www.giwersworld.org/bible/ot.phtml a6
 
Matt said:
If you have a journaling file system like EXT3 it has to save the journal. I
think the default is 5 seconds for ext3. You should have a journalling
filesystem if the release or update you use is less than five years old.
Click to expand...

I've now read up on this to the extent of my patience. Since I'm using
Fedora Core 5 I presume that what you say about my system is true.
But I didn't find any material on ext3 that could lead me to infer that
there was
this kind of high frequency cyclic activity, let alone how to modify
it.
 
I've now read up on this to the extent of my patience. Since I'm using
Fedora Core 5 I presume that what you say about my system is true.
But I didn't find any material on ext3 that could lead me to infer that
there was
this kind of high frequency cyclic activity, let alone how to modify
it.
Click to expand...

I have never looked into it and why do you care? There is no magic here. It
works. There is gnu version of the IBM system also. It does not hurt me in any
sense I have ever noticed. Even if it did the ext2 "forced disk check" on
booting would be a much greater annoyance.

I have no idea how this works. I presume you can at least recompile with a
greater time period. I agree that for us lowly home users that both journal size
and the update time are what we call extreme better suited for website hosters
and such.

If you really care, but I cannot see a reason why you would care, I presume
someone else here can point you to the docs and commandline to modify it.

But before jousting windmills make certain the disk activity is the exact same
number of seconds apart to see if my suggestion is correct.

Can you tell me why you would want to change it? Not trying to turn you off but
rather why? If you have some ancient hardware and/or more that five years old I
might see a reason.

--
The only good Zionist is a dead Zionist. It is a moral imperative to make
all Zionists into good Zionists.
-- The Iron Webmaster, 3695
nizkor http://www.giwersworld.org/nizkook/nizkook.phtml
Larry Shiff http://www.giwersworld.org/computers/newsagent.phtml a8
 
I've now read up on this to the extent of my patience. Since I'm using
Fedora Core 5 I presume that what you say about my system is true.
But I didn't find any material on ext3 that could lead me to infer that
there was
this kind of high frequency cyclic activity, let alone how to modify
it.
Click to expand...

Excuse me that was not an answer.

I _think_ the following will work but only if it is ext3.

# umount /hard/drive

# mount -t ext2 /hard/drive /mount/point

The edit /etc/fstab changing the ext3 to ext2. This will make it mount as ext2
next time you boot.

Then there is some other command where you can resize the journaling space to
something smaller and get back HD space. Forget what it is offhand.
 
Matt said:
(e-mail address removed) wrote:
Click to expand...
The edit /etc/fstab changing the ext3 to ext2. This will make it
mount as ext2
next time you boot.

Then there is some other command where you can resize the journaling space to
something smaller and get back HD space. Forget what it is offhand.
Click to expand...

Sounds like a simple-enough experiemt -- edit fstab to reboot in ext2
-- but how would this interact with journaling? If the reboot
failed...well then I'd really be up a creek. Anyway, it's mainly
curiosity on my part now. But I can't help pointing out that there's
been a tradeoff: hard drive motors used to shut down after a timeout
period as part of the energy-saving regime. Apparently this has been
traded off in favor of preparedness for...what? Lightning strike?
Terrorist attack?
 
Sounds like a simple-enough experiemt -- edit fstab to reboot in ext2
-- but how would this interact with journaling? If the reboot
failed...well then I'd really be up a creek.
Click to expand...

I said I THINK it will work. ext3 is simply a journaling version of ext2 but
otherwise identical so I think it will be safe. At least there have been no "you
idiot!" posts ... on this subject at least.
Anyway, it's mainly
curiosity on my part now. But I can't help pointing out that there's
been a tradeoff: hard drive motors used to shut down after a timeout
period as part of the energy-saving regime. Apparently this has been
traded off in favor of preparedness for...what? Lightning strike?
Terrorist attack?
Click to expand...

While linux systems don't go down much those lightning strikes in the form of
power line glitches for example often corrupt something on the HD. When
rebooting you get a "disk check forced" and it proceeds while you get some lunch
if just a 40GB HD. Some times you will get a message telling you do it manually
and then if you use the -y option while taking your after lunch nap. OK,
exaggerating but today with most drives over 100GB I would not be. Neither I nor
my son who works in the field have ever found anything recoverable in lost+found
so we see no reason not to use the -y option and then simply delete everything
in lost+found.

The journaling system recovers without forcing a diskcheck and you will get
that message during the boot process something like 'will recover during boot
process.'

If you are a company if the computer is a server everyone is waiting around for
the disk check with a non-journaling file system. If you are hosting service you
get angry customers and many fewer 9s on your availability bragging number.

As to power saving it means something on a laptop but it is around 30W on a
desktop, less than a KW-Hr per day. And if it does power down is it not annoying
to wait for it to get back up to speed?

--
Bush has announced himself as a Christian Zionist.
That means he is a traitor to America.
-- The Iron Webmaster, 3693
nizkor http://www.giwersworld.org/nizkook/nizkook.phtml
book review http://www.giwersworld.org/israel/willing-executioners.phtml a7
 
I have found 2 things that caused regular writes: the SAR service, and MRTG.
Having either of these installed sets a cron job that writes a log. Look
in /etc/cron.d for any files there and comment out their contents.
 
I have found 2 things that caused regular writes: the SAR service, and MRTG.
Having either of these installed sets a cron job that writes a log. Look
in /etc/cron.d for any files there and comment out their contents.
Click to expand...

And maybe others. That is why I first suggested timing the interval between
activity light blinks. I suggested (from memory not for certain) a five seconds
for journaling. He did not actually give a time but in replying implied that was
the interval.

BTW: An improperly configured mrtg emails root a failure notice. I had a small
drive on a machine where I never checked mail fill up. It is worth confirming it
is working properly.

--
Have you ever seen a pro-peace, anti-Israel Jew on TV? Why not?
What does that say about Jews?
-- The Iron Webmaster, 3689
nizkor http://www.giwersworld.org/nizkook/nizkook.phtml
Iraqi democracy http://www.giwersworld.org/911/armless.phtml a3
 
Matt said:
And maybe others. That is why I first suggested timing the interval between
activity light blinks. I suggested (from memory not for certain) a five seconds
for journaling. He did not actually give a time but in replying implied that was
the interval.
Click to expand...

The disk access LED flashes every two
seconds under Linux and every second
when I'm using Windows XP. Actually I
mentioned that in my original posting.
The flashes are brief and the intervals
between them so regular that it's obvious
that some timer-like process is controlling
them. This is what made me think that it
was hardware and so I also mentioned
which motherboard I had.

As for cron jobs under Linux: there are
aren't any! I was surprised to discover
that. I recall that with my Red Hat installation
of an eralier generation, several of them
were installed by default.

One thing I just tried this morning to
rule out the hardware at a really low
level was to boot my old Windows 98 recovery
floppy and then watch for a minute or
so. The disk-access LED was dark.
 
Your "problem" doesn't really seem like a problem at all, I
would say you're too worried, the machine at the office may have more
RAM than your home machine.

I would say your blinking LED is due to one of the following:

Journaling activity
Windows Indexing Service
Normal Paging activity
Buffer Flushes (Disk Cache)
Your AntiVirus/AntiSpyware Software

You must remember that booting with a win98 floppy does not mean that
the condition would be replicated, windows/linux are multi tasking
OS'es and there is alot more running than what you see on your
taskbar...

I'd just leave it alone, as long as it passes the tests (Virus/Spyware
ect..)
 
Back
Top