On Fri, 27 Mar 2015 14:12:24 +0800
Bret Busby <bret.bu...@gmail.com> wrote:

> On 27/03/2015, Bob Proulx <b...@proulx.com> wrote:
> > Bret Busby wrote:
> >> I have an external USB HDD connected to a system running Debian 6
> >> LTS.
> >
> > I don't really have any great contribution.  But since no one else
> > seems to have any good response I will contribute what I know.
> >
> > I have never had good luck with USB connected hard drives.  They
> > work for a while.  But then invariably they get dropped offline.
> > It might be 3-6 months between events.  But for me they just are
> > not reliable. In the past I have tried very hard to use them as
> > system disks.  Now I consider that something to avoid.
> >
> > I still use USB disks as "large floppies".  They are still great for
> > being large temporary data stores for holding and moving data
> > between machines.  But only when connected for short term use.
> > Reading your problems just reinforces this belief.
> >
> > [On the other hand USB network devices have been rock solid for me.
> > Meaning that while I avoid USB disks I actively use USB networking
> > on several machines to add additional NICs.  I am planning another
> > site using additional USB NICs.  It is probably hardware dependent
> > but they have been working great for me regardless of the opposite
> > for disks. And I have three sites using USB sound cards very
> > robustly.]
> >
> >> I have tried to transfer data from the desktop intenal HDD, to the
> >> external USB HDD.
> >>
> >> The file manager shows as being "Nautilus 2.30.1".
> >
> > I read by your message that you are a graphical desktop user.
> > That's fine.  But for transfering large amounds of data the command
> > line tools such as rsync are the best in class.  I wouldn't even
> > consider trying to use nautilus or other graphical file managers
> > for this type of task.  I would highly recommend using rsync.  Even
> > if for you it means a stretch to get off of the mouse and over to
> > the keyboard.
> >
> > The best advantage of tools such as rsync is that it is
> > interruptable and restartable with a minimum of lost effort.  I may
> > be 1G into a 3G transfer and want to stop it, change something, and
> > restart it again.  With a normal copy that would mean copying the
> > original data again.  With rsync it means it will examine what needs
> > to be done and be able to continue the copy using the already
> > transfered data as done and moving forward.
> >
> >   rsync -avP /from/here/dir-or-file /to/there/dir/
> >
> >> From time to time, as in this instance, I forget (until too late)
> >> that Debian 6 can not cope with transferring data more than about
> >> 1GB at a time; in this instance, I had tried to transfer about
> >> 3GB, to make room in my /home partition.
> >
> > Knowing how flaky USB disks interfaces tend to be I think this is
> > most likely a hardware problem.  Doesn't change your situation.
> > But I think it blames the right thing to blame.
> >
> > In any case I have definitely copied gigs and gigs of data to and
> > from USB disks.  It can be very good to make large data sets
> > portable on a portable USB drive.  My complaints usually happen
> > after the disk has been in active use as a system device for a
> > month and then it goes offline.
> >
> >> The transfer had seized up, after transferring about 1.1GB of the
> >> 3.2GB that I had tried to ransfer, so, after a couple of days of it
> >> apparently doing nothing, I stopped it, and, as the system monitor
> >> showed a system load of around 43 (whatever that means - if it was
> >> as a percentage of system capacity, it could be more meaningful,
> >> to me). The system monitor currently shows a system load average
> >> of about 35.
> >
> > Let me give a short explanation of system load.  Which is almost
> > impossible to say briefly so forgive me in advance for leaving out
> > important parts, saying half of it wrong, and still saying too much.
> > The concept is the important part here.
> >
> > First there is no set capacity.  There isn't a cap such as 5 or 10
> > or 100.  Therefore there isn't a way to say what percentage of your
> > system is being used by any particular system load.  But it is an
> > important indicator of system status and health.  A load of 35 or 43
> > are both very high system loads!
> >
> > The operating system process scheduler schedules processes to run.
> > A process ready to run is queued into the run queue.  If the process
> > is calculating PI to a zillion decimal places then it is going to
> > use 100% of the cpu until it has consumed its time slice and
> > suspended to give the next process time to run.  If there are no
> > other proceses then the cpu will be given back to this process and
> > the cpu will continue to be 100% utilized forever.
> >
> > But what about processes reading and writing to the disk drive or
> > network?  In computer speed spinning disk drives are slow.  In
> > computer speeds networks are slow.  Web servers are slow.  Say that
> > your web browser sends an http GET request to a web site.  It then
> > must wait for the response.  Your web browser is ready to run.  But
> > it can't.  It is waiting for external events return.  It is
> > "blocked" waiting for I/O.  While it is waiting the OS will
> > schedule another process to run.  If there is another process ready
> > it will get cpu. It may also be waiting for the disk drive to spin
> > and return data.  Or spin and complete a data write.
> >
> > The OS will spin through the run queue running any process that is
> > ready to run.  Some will run for a short time and then perform a
> > read or a write to an I/O device.  That will cause them to stop
> > running while their I/O request is processed.  Other processes such
> > as those running protein folding research will consume all cpu
> > running until the OS suspends them.  The OS manages the run queue
> > and keeps things moving through the cpu using a variety of
> > algorithms to make best use of the cpu.  A system running a lot of
> > processes where all of the processes are using 100% cpu will be
> > very busy and feel very slow.  A system running a lot of processes
> > where most of them are waiting for some I/O to complete will be
> > mostly idle because it will be mostly waiting and will feel very
> > responsive.  You might not even notice it is doing so many things.
> >
> > I hope this explains the run queue in the OS and how it is used.
> > The run queue is the load average you are seeing.  Those numbers 43
> > and 35 are the number of processes in the OS run queue.  It says
> > that you have forty processes in the run queue.  Wow!  In a typical
> > desktop system the load average will hover near zero.  That your
> > system is so large says something is blocking those processes from
> > completing. Your system may be very responsive if they are all
> > waiting, if they are all "blocked waiting for I/O".  Or it may be
> > really bogged down if they are all crunching something on the cpu.
> > Or bogged down if they are consuming too much memory and your
> > system swaps.  A high load average indicates something is blocking
> > those processes from completing and they are stacking up in the
> > queue.  When airplanes can't land at an airport fast enough then
> > air traffic control stacks them up in the holding patterns.  That
> > is what is being indicated by your high load average.  There isn't
> > a maximum and therefore there isn't a percentage of capacity.
> >
> > If you see a high load average such as that run a 'ps' command to
> > look to see what processes are running.  Sorry I don't know how to
> > do that from a graphical desktop.  But in the command line shell I
> > would normally run:
> >
> >   ps -efH | less   # Or the bsd way: ps aux | less
> >
> > Then look to see what is running.  If you have a load of 40 then
> > there will be at least 40 unusual processes there.  Maybe that is 40
> > nautilus processes stacked up trying to talk to the USB.  I think
> > that likely.  That would definitely indicate a system problem
> > talking to the USB device.
> >
> > You may ask, "What is an unusual process?"  That is a hard one to
> > answer.  You should look at a normally running system routinely and
> > learn what is normally running.  Then later when things are not
> > running normally you will recognize patterns that are different.
> > For example you would probably not normally see nautilus running.
> > Or if you have it open I expect you would see one process.  But if
> > you were to see 40 nautilus procesess then you would know that was
> > unusual and likely an indicator of a problem.
> >
> > What you would do with that information depends upon the
> > information. I would try to figure out why those processes are
> > stacking up.  I would try to clear them.  That might be by removing
> > the USB.  Or by killing processes with kill.  Or other things
> > depending upon the situation.  But something is wrong and it is
> > good to know what.
> >
> > I don't really know what to say about the rest of your problems and
> > situation.  But I hope this small piece was helpful for this small
> > part of it.
> >
> > Bob
> >
> 
> 
> Hello.
> 
> Thank you for your advice.
> 
> I have, on the internal HDD, two data partitions, each of which,
> should be able to take the data stored on  the USB HDD. However, I
> need to make them writable for me as a user, to be able to copy the
> data. They show as being owned by root, and so, are not witable for me
> as a user, when I mount them, using the disk utility. I will have to
> investigate changing the permissions. That should be relatively easy.
> 
> And, I need to find an alternative, for backing up data.
> 
> The iomega zip drives, seemed to be okay, many years ago, when 100MB
> was all that a person would be likely to have as data on a HDD, but,
> nowadays, it is all, so much bigger.
> 
> And, I do know that HDD's fail, eventually. They may take different
> periods of time, but, like animals (including humans), most things
> die, eventually, and, in the case of HDD's, that means losing all
> stored data, unless it is securely "backed up".
> 
> So, that is something else that I need to investigate.
> 
> Many people (including me) have lost data, due to it not being backed
> up, and thence, lose the data, when something goes wrong. And, I have
> heard tell of a PhD student, who lost his finalised thesis, due to a
> HDD or system failure, the thesis not being backed up. It was not a
> case of "the dog ate my homework", so much as "the computer ate my
> homework".
> 
> A few years ago, in a move operation ( in that case, moving data from
> one computer to another), it all went wrong, and I lost about 20 years
> of research, in a particular, and, to me, important, area, because it
> was not backed up. I was going from a system with one version of
> Debian, to a new system with a newer version of Debian,something like
> between 3.x to 4, or, 4 to 5, I think.
> 
> I have since, learnt (when I remember it) to use copy -> paste ->
> check all data has been successfully copied -> delete original data at
> source, instead of move.
> 
> An issue then, now, is finding a (relatively) safe medium, to use for
> backing up data.
> 

What I use are bluray discs. You can find a burner relatively cheaply,
and the discs themselves aren't all that expensive if you weigh it
against what your data is worth. The discs themselves hold 25GB, which
is enough for me, it might not be for you, depending on how much data
you need to back up. I also have a couple of rewritable discs, which
are more expensive and burn much slower, but I don't really know how
reliable these are and how many times you can burn them.

My backups are taken so frequently that if a disc fails to restore, I
won't lose much data by going another disc back in the "chain". I also
have "live" backups on a hard disk, in case something is accidentally
deleted or corrupted. Best practice is to not rely on a single backup.

Steve Litt (on this list - hi, Steve! Hope you don't mind me posting
this.) has a very helpful guide on backing up to bluray:

http://www.troubleshooters.com/linux/blu-ray-backup.htm

As Bob suggested, rsync really is the way to go. There is a GUI backup
program that's called backintime that uses rsync, I use it frequently
as it is fairly quick and easy to use, and gives me the control I need.
You might want to look into that if you would prefer not to use the
shell.

I hope this is helpful, feel free to ask if there is anything you need
to know.

Good luck!

Petter

-- 
"I'm ionized"
"Are you sure?"
"I'm positive."

Attachment: pgp_TFcKr9PtC.pgp
Description: OpenPGP digital signature

Reply via email to