[gentoo-user] Hard drive storage questions

[gentoo-user] Hard drive storage questions

[Dale]

Howdy,

I have a 3TB hard drive that I use for my /home partition.  I'm going to
be having to expand this before to long, lots of videos on there.  The
4TB is a bit pricey and I would end up having to expand that to before
to long.  So, I got to thinking, why not buy another 3TB drive and just
add that which would double my space.  I use LVM by the way.  I may try
BTFS, (sp?).   Either way, adding a drive shouldn't be to much of a
problem.

On one hand, adding a drive would double my space.  It would also spread
out my stuff in the event a drive failed.  On the other hand, one more
drive to have spinning that could fail too.  These large drives makes me
wonder sometimes.

What do you guys, gals too, think about this?  Just add a drive or buy a
larger drive and move things over?  Or is this a six of one and half
dozen of the other thing?

Dale



P. S.


Filesystem      	 Size      Used     Avail    Use%     Mounted on
/dev/mapper/Home2-Home2  2.7T     1.8T      945G      66%     /home

Re: [gentoo-user] Hard drive storage questions

[Francisco Ares]

[-- Attachment #1: Type: text/plain, Size: 1751 bytes --]

2015-04-28 5:39 GMT-03:00 Dale <rdalek1967@gmail.com>:

> Howdy,
>
> I have a 3TB hard drive that I use for my /home partition.  I'm going to
> be having to expand this before to long, lots of videos on there.  The
> 4TB is a bit pricey and I would end up having to expand that to before
> to long.  So, I got to thinking, why not buy another 3TB drive and just
> add that which would double my space.  I use LVM by the way.  I may try
> BTFS, (sp?).   Either way, adding a drive shouldn't be to much of a
> problem.
>
> On one hand, adding a drive would double my space.  It would also spread
> out my stuff in the event a drive failed.  On the other hand, one more
> drive to have spinning that could fail too.  These large drives makes me
> wonder sometimes.
>
> What do you guys, gals too, think about this?  Just add a drive or buy a
> larger drive and move things over?  Or is this a six of one and half
> dozen of the other thing?
>
> Dale
>
>
>
> P. S.
>
>
> Filesystem               Size      Used     Avail    Use%     Mounted on
> /dev/mapper/Home2-Home2  2.7T     1.8T      945G      66%     /home
>
>
>
>
I've already had my /home partition filled up, and it was just one of the
partitions on the only hard drive inside this machine.

Nowadays I have 3 more hard drives on this same machine, which makes 4
drives in total. None of them so big, though, the newest one is 1TB.

Now, talking about redundancy, if you worry about many - or all - of your
files, perhaps you should consider a RAID. Although expensive, it makes
one's mind more relaxed ;-) .  Basically, what I have seen is that many
people worry about backing up files when at work, but completely forget
about it on personal computers.

Best regards, and good luck!
Francisco

[-- Attachment #2: Type: text/html, Size: 2419 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Alan McKinnon]

On 28/04/2015 10:39, Dale wrote:
> Howdy,
> 
> I have a 3TB hard drive that I use for my /home partition.  I'm going to
> be having to expand this before to long, lots of videos on there.  The
> 4TB is a bit pricey and I would end up having to expand that to before
> to long.  So, I got to thinking, why not buy another 3TB drive and just
> add that which would double my space.  I use LVM by the way.  I may try
> BTFS, (sp?).   Either way, adding a drive shouldn't be to much of a
> problem.
> 
> On one hand, adding a drive would double my space.  It would also spread
> out my stuff in the event a drive failed.  On the other hand, one more
> drive to have spinning that could fail too.  These large drives makes me
> wonder sometimes.
> 
> What do you guys, gals too, think about this?  Just add a drive or buy a
> larger drive and move things over?  Or is this a six of one and half
> dozen of the other thing?
> 
> Dale
> 
> 
> 
> P. S.
> 
> 
> Filesystem      	 Size      Used     Avail    Use%     Mounted on
> /dev/mapper/Home2-Home2  2.7T     1.8T      945G      66%     /home
> 
> 
> 


When you're up into the TB range you run a higher risk of losing data
than with disks of a few 100 GB simply because it's bigger and there are
more bits that can flip [1].

When you use only LVM for this and nothing else, you have a high risk of
losing everything if one disk fails. Why? Because LVM decides itself
which extent it will put data on. Maybe a whole file is on one disk,
maybe it's spread across two, because the software is designed so that
you don't have to be concerned with that. The only thing that LVM does
is expand your storage space as a single volume and make it easier to
shuffle things around without having to backup/repartition/restore.

The best solution for you depends on what you need and what you have. If
your disks are full of YouTube videos that you can easily download again
(or stream), maybe you don't care too much. Precious family photos that
can't be replaced? You need to care a lot.

Personally, I like the ZFS approach and do it all in software, catching
errors that RAID misses.
RAID is also an option - 1:1 mirroring works great if you are much more
concerned about data than about cost.

There is no general advice in this area[2], the trick is to understand
the various technologies, fully understand your own needs and budget,
then plan accordingly.



[1] All things being equal that is. A 3TB disk is probably not really
the same as a 500G disk, just bigger. It's safe to assume that disk
manufacturers pat attention to error rates etc and improve their
products over time to make them more reliable. As to by how much - I
don't know.

[2] There is however a vendor's desire to maximize their profit while
still leaving you with warm and fuzzies </sarcasm>

-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Tue, Apr 28, 2015 at 4:39 AM, Dale <rdalek1967@gmail.com> wrote:
>
> What do you guys, gals too, think about this?  Just add a drive or buy a
> larger drive and move things over?  Or is this a six of one and half
> dozen of the other thing?
>

Generally I buy drives at the sweet spot in cost/capacity, so that is
about 3TB last time I checked (for spinning disks).

I ALWAYS use RAID or full backups of some kind.  RAID isn't really a
substitute for backups, but I use it as such for low-priority data
such as mythtv recordings or re-generatable data.  Right now I'm
running on mirrored btrfs with a full backup to ext4 (since btrfs is
living dangerously).  I'm actually getting tight on space and debating
dropping the full backups for lower-priority data, which would free up
a 3TB drive to add to the btrfs array.  Long-term I'd prefer to move
to raid5 which is much more efficient, but I wouldn't recommend doing
that on btrfs yet - it is very immature.

raid5 on mdadm and lvm with ext4 is very mature, and is probably your
most space-efficient option with some level of redundancy.  With large
arrays having raid6 isn't a bad idea these days - it takes a lot of
time to recover a failure.  However, if you have a single drive today
there is no way to add only a single disk and get both more space and
redundancy at the same time.  If you want more space and only want to
buy one drive, then you're stuck with just simple lvm and if a drive
fails you're going to lose a lot of stuff.

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1391 bytes --]

On Tue, 28 Apr 2015 17:01:49 +0200, Alan McKinnon wrote:

> When you use only LVM for this and nothing else, you have a high risk of
> losing everything if one disk fails. Why? Because LVM decides itself
> which extent it will put data on. Maybe a whole file is on one disk,
> maybe it's spread across two, because the software is designed so that
> you don't have to be concerned with that. The only thing that LVM does
> is expand your storage space as a single volume and make it easier to
> shuffle things around without having to backup/repartition/restore.

An alternative is to create a new volume group on the new disk and mounts
PVs at various points in your home directory. That way you get the extra
space and much of the flexibility without the risk of a failure on a
single drive taking out data on both. However, if you are concerned about
data loss, you should be using RAID t a minimum, preferably with an error
detecting filesystem.

> Personally, I like the ZFS approach and do it all in software, catching
> errors that RAID misses.

The same is also  possible with BTRFS, including built in RAID. RAID5 in
btrfs is expermiental, but its RAID1 is like RAID5 in some ways, such as
giving the capacity of n-1 disks and tolerating a single disk failure.


-- 
Neil Bothwick

PC DOS Error #04: Out of disk space. Delete Windows? (Y)es (H)ell yes!

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Tue, Apr 28, 2015 at 11:24 AM, Neil Bothwick <neil@digimed.co.uk> wrote:
> The same is also  possible with BTRFS, including built in RAID. RAID5 in
> btrfs is expermiental, but its RAID1 is like RAID5 in some ways, such as
> giving the capacity of n-1 disks and tolerating a single disk failure.
>

btrfs raid5 is still fairly experimental (though now it supports
recovery) and works more-or-less how you'd expect raid5 to work.
Raid1 on btrfs gives you the capacity of n/2 and not n-1 disks, as you
would expect.  It does allow disks to be of different size, in which
case it gives you up to n/2 capacity (and usually more than you'd get
with traditional raid1 - it tries to fill the largest disks first so
3+1+1+1 TB will give you 3TB of storage, not 2TB as you'd get with
mdadm raid1).

Here is a btrfs raid1:
df -h
Filesystem                   Size  Used Avail Use% Mounted on
/dev/sdh2                    3.2T  2.7T  477G  86% /data

btrfs fi df /data
Data, RAID1: total=2.93TiB, used=2.65TiB
System, RAID1: total=32.00MiB, used=472.00KiB
Metadata, RAID1: total=16.00GiB, used=14.08GiB
GlobalReserve, single: total=512.00MiB, used=0.00B

btrfs fi sho /data
Label: 'datafs'  uuid: cd074207-9bc3-402d-bee8-6a8c77d56959
        Total devices 5 FS bytes used 2.67TiB
        devid    1 size 2.73TiB used 2.63TiB path /dev/sdh2
        devid    2 size 931.32GiB used 832.03GiB path /dev/sda2
        devid    3 size 931.32GiB used 834.00GiB path /dev/sde2
        devid    4 size 931.32GiB used 832.00GiB path /dev/sdd2
        devid    5 size 931.32GiB used 833.00GiB path /dev/sdb2


2.7TiB of data is stored on the array, which has nearly exhausted the
space of 6.4TiB of drives (or 7TB).  There are ~500GiB free, which
would let me store ~250GiB of data.

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 790 bytes --]

On Tue, 28 Apr 2015 13:38:55 -0400, Rich Freeman wrote:

> > The same is also  possible with BTRFS, including built in RAID. RAID5
> > in btrfs is expermiental, but its RAID1 is like RAID5 in some ways,
> > such as giving the capacity of n-1 disks and tolerating a single disk
> > failure. 
> 
> btrfs raid5 is still fairly experimental (though now it supports
> recovery) and works more-or-less how you'd expect raid5 to work.
> Raid1 on btrfs gives you the capacity of n/2 and not n-1 disks,

You're right, I was clearly confused (an oxymoron?) when I wrote that.

So RAID1 gives less capacity than RAID5 on BTRFS, but it is stable (in
btrfs terms).


-- 
Neil Bothwick

This universe is sold by mass, not by volume.
Some expansion may have occurred during shipment

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Tue, Apr 28, 2015 at 2:11 PM, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Tue, 28 Apr 2015 13:38:55 -0400, Rich Freeman wrote:
>
>> > The same is also  possible with BTRFS, including built in RAID. RAID5
>> > in btrfs is expermiental, but its RAID1 is like RAID5 in some ways,
>> > such as giving the capacity of n-1 disks and tolerating a single disk
>> > failure.
>>
>> btrfs raid5 is still fairly experimental (though now it supports
>> recovery) and works more-or-less how you'd expect raid5 to work.
>> Raid1 on btrfs gives you the capacity of n/2 and not n-1 disks,
>
> You're right, I was clearly confused (an oxymoron?) when I wrote that.
>
> So RAID1 gives less capacity than RAID5 on BTRFS, but it is stable (in
> btrfs terms).
>

Correct.

For drives of identical size and not using compression, I'd expect
space use on btrfs to be equivalent to the same raid level on
mdadm+lvm+ext4.  With mixed drives you will potentially get more space
on btrfs, and compression will of course get you more space.

As far as data security goes there is a tradeoff.  Btrfs is still
immature and I seem to have issues with it 1-2 times per year (but
I've yet to have unrecoverable data loss).  On the other hand, btrfs
does do full data checksumming which means you're less likely to lose
data due to issues with the physical storage than with mdadm - as with
zfs it always checks the checksum and will recover from another disk
if possible, and in the event of raid disparity it always knows which
(if any) of the copies is right.

I'm hopeful that at some point I'll be able to recommend it without
reservation.  Right now, that isn't entirely the case.  I'm still
patching the 3.18 kernel series so that it actually mounts my root
partition when whatever is causing it to panic (probably also btrfs)
does so (the patch is in the queue, but hasn't made it to 3.18 yet for
some reason - I believe it has been in the other stable series for a
release or two now).

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1066 bytes --]

On Tue, 28 Apr 2015 14:31:23 -0400, Rich Freeman wrote:

> >> btrfs raid5 is still fairly experimental (though now it supports
> >> recovery) and works more-or-less how you'd expect raid5 to work.
> >> Raid1 on btrfs gives you the capacity of n/2 and not n-1 disks,  
> >
> > You're right, I was clearly confused (an oxymoron?) when I wrote that.
> >
> > So RAID1 gives less capacity than RAID5 on BTRFS, but it is stable (in
> > btrfs terms).
> >  
> 
> For drives of identical size and not using compression, I'd expect
> space use on btrfs to be equivalent to the same raid level on
> mdadm+lvm+ext4.

That's only true for RAID1 when you have 2 drives, with more drives btrfs
gives more space.

3 x 2TB drives give 2TB on MD RAID1
3 x 2TB drives give 3TB on btrfs RAID1

Although you will get slightly less usable space with btrfs because of
the space used for metadata, but I'm not sure how significant that is.


-- 
Neil Bothwick

QOTD:
	The only easy way to tell a hamster from a gerbil is that the
	gerbil has more dark meat.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: Hard drive storage questions

[walt]

On 04/28/2015 08:24 AM, Neil Bothwick wrote:
> On Tue, 28 Apr 2015 17:01:49 +0200, Alan McKinnon wrote:
 
>> Personally, I like the ZFS approach and do it all in software, catching
>> errors that RAID misses.
> 
> The same is also  possible with BTRFS,

I have the impression (without knowing what I'm talking about) that BTRFS
was created to be just like ZFS, minus the software licensing problems.

Is my impression right or wrong?

Re: [gentoo-user] Re: Hard drive storage questions

[Rich Freeman]

On Tue, Apr 28, 2015 at 6:02 PM, walt <w41ter@gmail.com> wrote:
> On 04/28/2015 08:24 AM, Neil Bothwick wrote:
>> On Tue, 28 Apr 2015 17:01:49 +0200, Alan McKinnon wrote:
>
>>> Personally, I like the ZFS approach and do it all in software, catching
>>> errors that RAID misses.
>>
>> The same is also  possible with BTRFS,
>
> I have the impression (without knowing what I'm talking about) that BTRFS
> was created to be just like ZFS, minus the software licensing problems.
>
> Is my impression right or wrong?

Kinda.  Same sort of idea, and the licensing obviously has a big part in it.

The underlying designs are different, which means that when fully
mature each will likely have different pros/cons, which is great since
we can all pick/choose what we need.

The other difference is that ZFS is targeted more at enterprise /
large-scale use, and btrfs is targeted more as a general-purpose
filesystem that you might use on a single-disk PC.  That isn't to say
that either can't be used in either situation, but you can definitely
see where there has been more focus in feature development.  For
example, with zfs you can not only have large pools of drives, but you
can also bind them into smaller redundancy pools.  So, you can have 10
"raid6" arrays bound together which ensures that the scale of rebuilds
is limited while giving you a common pool of space.  On the other
hand, with btrfs you can have a 3-disk raid5 and turn it into a 4-disk
raid5 without having to copy/restore all your data (or you could turn
a 3-disk raid1 into a 4-disk raid5, and even switch halfway so that
half your data is in raid1 mode and half in raid5).  That is the sort
of thing that is handy in a small PC where you don't just have stacks
of disks lying around to build fresh new arrays from, but less
important for a big enterprise SAN where you don't need to add one
disk at a time to a 40-disk storage unit.

I'm sure many features exclusive to either btrfs or zfs will
eventually make their way to the other.  However, their differing
focuses make it likely that some features will mature faster than
others.

And of course btrfs has been taking a fairly long time to mature - it
just doesn't seem like as serious of an enterprise-y project.  But,
neither is Gentoo.  :)

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Alan McKinnon]

On 28/04/2015 17:24, Neil Bothwick wrote:
> On Tue, 28 Apr 2015 17:01:49 +0200, Alan McKinnon wrote:
> 
>> When you use only LVM for this and nothing else, you have a high risk of
>> losing everything if one disk fails. Why? Because LVM decides itself
>> which extent it will put data on. Maybe a whole file is on one disk,
>> maybe it's spread across two, because the software is designed so that
>> you don't have to be concerned with that. The only thing that LVM does
>> is expand your storage space as a single volume and make it easier to
>> shuffle things around without having to backup/repartition/restore.
> 
> An alternative is to create a new volume group on the new disk and mounts
> PVs at various points in your home directory. That way you get the extra
> space and much of the flexibility without the risk of a failure on a
> single drive taking out data on both. However, if you are concerned about
> data loss, you should be using RAID t a minimum, preferably with an error
> detecting filesystem.


I've used that scheme myself in the past. You do get the increased space
but you don't get much in the way of flexibility. And it get COMPLICATED
really quickly.

To get around the situation of one drive almost full and the other
having lots of space, folks often use symlinked directories, which you
forget about and no-one else can figure out what you did...

It all smacks of the old saw:

For any non-trivial problem, there is always at least one solution that
is simple, elegant, and wrong.



-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: Hard drive storage questions

[Alan McKinnon]

On 29/04/2015 00:02, walt wrote:
> On 04/28/2015 08:24 AM, Neil Bothwick wrote:
>> On Tue, 28 Apr 2015 17:01:49 +0200, Alan McKinnon wrote:
>  
>>> Personally, I like the ZFS approach and do it all in software, catching
>>> errors that RAID misses.
>>
>> The same is also  possible with BTRFS,
> 
> I have the impression (without knowing what I'm talking about) that BTRFS
> was created to be just like ZFS, minus the software licensing problems.
> 
> Is my impression right or wrong?


As with all things, it's probably more complicated than that :-)

I personally think that ZFS (from Sun Microsystems) and BTRFS (from
Oracle) were originally convergent solutions to the same problem, much
like Gnome and KDE both try solve the desktop problem. ZFS started out
in the Solaris world, and BTRFS in the Oracle-cloned-Red-Hat world, so
there is that difference.

Then Oracle bought Sun and now Oracle "owns" both codebases, so who
knows what's going in internally at that corporation wrt modern filesystems.

ZFS licensing is a problem that should not exist. AFAIK, Sun owned the
entire codebase and used their own license. Oracle owns it now, so there
doesn't seem to be anything stopping Oracle from releasing the whole
thing under multiple licenses, making the problem go away.


-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1809 bytes --]

On Wed, 29 Apr 2015 08:13:41 +0200, Alan McKinnon wrote:

> > An alternative is to create a new volume group on the new disk and
> > mounts PVs at various points in your home directory. That way you get
> > the extra space and much of the flexibility without the risk of a
> > failure on a single drive taking out data on both. However, if you
> > are concerned about data loss, you should be using RAID at a minimum,
> > preferably with an error detecting filesystem.  
> 
> I've used that scheme myself in the past. You do get the increased space
> but you don't get much in the way of flexibility. And it get COMPLICATED
> really quickly.

It certainly can, but for a simple two drive home system it shouldn't get
out of hand. However, it does avoid the "one disk errors kills two
disks' data" problem.

> To get around the situation of one drive almost full and the other
> having lots of space, folks often use symlinked directories, which you
> forget about and no-one else can figure out what you did...

I wasn't suggesting symlinks, just LVs mounted at appropriate points. It
rather depends on the spread of Dale's data. If he just needs extra space
for his videos, he could get a new drive and mount it at ~/videos.

> It all smacks of the old saw:
> 
> For any non-trivial problem, there is always at least one solution that
> is simple, elegant, and wrong.

:-)

I consider what I suggested somewhat simple but far from elegant. Often
though, it's a lot less work in the long run to go for the initially more
complex solution. If Dale is worried about the likelihood of disk
failure, he really should be using RAID - either MDRAID under LVM or one
of the next-gen filesystems.


-- 
Neil Bothwick

... Taglines: and How They Affect Women. Next On Oprah.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: Hard drive storage questions

[Grant Edwards]

On 2015-04-29, Alan McKinnon <alan.mckinnon@gmail.com> wrote:

Regarding ZFS licensing problems:

> [...] there doesn't seem to be anything stopping Oracle from [...]
> making the problem go away.

In my rather limited experience with Oracle, "making the problem go
away" never seemed to be high on their agenda for anything they were
involved in.

-- 
Grant Edwards               grant.b.edwards        Yow! Now we can become
                                  at               alcoholics!
                              gmail.com            

Re: [gentoo-user] Hard drive storage questions

[Dale]

Dale wrote:
> Howdy,
>
> <<  SNIP >>
>
> Dale
>
>
>
> P. S.
>
>
> Filesystem      	 Size      Used     Avail    Use%     Mounted on
> /dev/mapper/Home2-Home2  2.7T     1.8T      945G      66%     /home
>
>
>


Well, I read replies a few times and I think it is best to just add a
new drive.  Heck, I've already had a 3TB drive to fail.  Anyway, I also
need to look into some sort of backup system.  I used to do this with
DVDs but with this much "stuff", that just isn't a good idea, not to
mention that DVDs have their own issues.   I may take a peek into a RAID
setup since really, that is about the best if not only way to do it. 

Thanks all for the replies.

Dale

:-)  :-) 

Re: [gentoo-user] Hard drive storage questions

[Dale]

[-- Attachment #1: Type: text/plain, Size: 2687 bytes --]

Neil Bothwick wrote:
> On Wed, 29 Apr 2015 08:13:41 +0200, Alan McKinnon wrote:
>
>>> An alternative is to create a new volume group on the new disk and
>>> mounts PVs at various points in your home directory. That way you get
>>> the extra space and much of the flexibility without the risk of a
>>> failure on a single drive taking out data on both. However, if you
>>> are concerned about data loss, you should be using RAID at a minimum,
>>> preferably with an error detecting filesystem. 
>>
>> I've used that scheme myself in the past. You do get the increased space
>> but you don't get much in the way of flexibility. And it get COMPLICATED
>> really quickly.
>
> It certainly can, but for a simple two drive home system it shouldn't get
> out of hand. However, it does avoid the "one disk errors kills two
> disks' data" problem.

Yea, right now, I'm only using two drives.  One for the OS and one for
/home.  I have a third drive but it isn't in use.  I'm thinking about
moving everything but the videos to that drive, 750GB, and leave just
the videos on the large 3TB drive.  It'll free up a *little* space too.


>
>
>> To get around the situation of one drive almost full and the other
>> having lots of space, folks often use symlinked directories, which you
>> forget about and no-one else can figure out what you did...
>
> I wasn't suggesting symlinks, just LVs mounted at appropriate points. It
> rather depends on the spread of Dale's data. If he just needs extra space
> for his videos, he could get a new drive and mount it at ~/videos.

The bulk of the space is used by the videos.  It's everything from TV
shows to movies to youtube howtos.  I'm using roughly 1.8TB on the drive
and the videos take up roughly 1.7TB of that space.  My camera pics only
use 21GBs of space.  Rest is basically a rounding error.  :/

>
>> It all smacks of the old saw:
>>
>> For any non-trivial problem, there is always at least one solution that
>> is simple, elegant, and wrong.
>
> :-)
>
> I consider what I suggested somewhat simple but far from elegant. Often
> though, it's a lot less work in the long run to go for the initially more
> complex solution. If Dale is worried about the likelihood of disk
> failure, he really should be using RAID - either MDRAID under LVM or one
> of the next-gen filesystems.
>
>

I really do need to set up RAID at least for some stuff that I may not
be able to get back.  Some videos I have are no longer available.   What
I wish, I had a second puter in a outbuilding that I could copy to over
ethernet or something.  May help in the event of a house fire etc.

Dale

:-)  :-)


[-- Attachment #2: Type: text/html, Size: 3790 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1365 bytes --]

On Mon, 04 May 2015 02:39:10 -0500, Dale wrote:

> > I wasn't suggesting symlinks, just LVs mounted at appropriate points.
> > It rather depends on the spread of Dale's data. If he just needs
> > extra space for his videos, he could get a new drive and mount it at
> > ~/videos.
> 
> The bulk of the space is used by the videos.  It's everything from TV
> shows to movies to youtube howtos.  I'm using roughly 1.8TB on the drive
> and the videos take up roughly 1.7TB of that space.  My camera pics only
> use 21GBs of space.  Rest is basically a rounding error.  :/

You need to separate those anyway, for backup purposes. Anything you
downloaded, you can usually download again, so you only need a list of
the files to be able to find them again.

On the other hand, you photos are irreplaceable and need to be backed up.
 
> I really do need to set up RAID at least for some stuff that I may not
> be able to get back.  Some videos I have are no longer available.

RAID is not a backup solution.

> What
> I wish, I had a second puter in a outbuilding that I could copy to over
> ethernet or something.  May help in the event of a house fire etc.

You have, it's called Amazon S3 :) It's a lot cheaper than a second
computer, and a lot more reliable.
 

-- 
Neil Bothwick

Two rights don't make a wrong, they make an airplane.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 799 bytes --]

On Monday 04 May 2015 08:46:26 Neil Bothwick wrote:
> On Mon, 04 May 2015 02:39:10 -0500, Dale wrote:

> > I really do need to set up RAID at least for some stuff that I may not
> > be able to get back.  Some videos I have are no longer available.
> 
> RAID is not a backup solution.

Not only RAID 1 isn't a back up solution, because it offers temporary 
redundancy rather than diverse protection, but under certain scenarios you 
have a much higher chance of losing your data when the first drive fails.  If 
you bought two (or more) drives at the same time and built a RAID from them, 
their failure performance due to same construction and age could be quite 
similar.  On many occasions your last healthy drive fails, just as you try to 
rebuild the RAID.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Dale]

[-- Attachment #1: Type: text/plain, Size: 2089 bytes --]

Neil Bothwick wrote:
> On Mon, 04 May 2015 02:39:10 -0500, Dale wrote:
>
>>> I wasn't suggesting symlinks, just LVs mounted at appropriate points.
>>> It rather depends on the spread of Dale's data. If he just needs
>>> extra space for his videos, he could get a new drive and mount it at
>>> ~/videos.
>>
>> The bulk of the space is used by the videos.  It's everything from TV
>> shows to movies to youtube howtos.  I'm using roughly 1.8TB on the drive
>> and the videos take up roughly 1.7TB of that space.  My camera pics only
>> use 21GBs of space.  Rest is basically a rounding error.  :/
>
> You need to separate those anyway, for backup purposes. Anything you
> downloaded, you can usually download again, so you only need a list of
> the files to be able to find them again.
>
> On the other hand, you photos are irreplaceable and need to be backed up.

Well, some videos aren't available either.  I'd hate to know I had to
find some of the ones that are available.  Some take some diggin IF I
can even remember some of them.

My pics I backup to DVDs, two sets just in case.  I keep those in a
outbuilding.  If everything here burns, I'm likely gone anyway.


>
> 
>> I really do need to set up RAID at least for some stuff that I may not
>> be able to get back.  Some videos I have are no longer available.
>
> RAID is not a backup solution.

True but at least it would help if a drive fails.  I've been there a
couple times.

>
>
>> What
>> I wish, I had a second puter in a outbuilding that I could copy to over
>> ethernet or something.  May help in the event of a house fire etc.
>
> You have, it's called Amazon S3 :) It's a lot cheaper than a second
> computer, and a lot more reliable.
> 
>


My internet is way to slow for that.  It would take weeks maybe a month
to upload all this stuff.  I have DSL but it is the basic package.  If I
were on cable or had a real fast DSL, maybe.  Thing is, I really don't
want some of my stuff on the internet anyway.  ;-)

I'll come up with something tho.

Dale

:-)  :-)


[-- Attachment #2: Type: text/html, Size: 3087 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Dale]

[-- Attachment #1: Type: text/plain, Size: 1018 bytes --]

Mick wrote:
> On Monday 04 May 2015 08:46:26 Neil Bothwick wrote:
>> On Mon, 04 May 2015 02:39:10 -0500, Dale wrote:
>
>>> I really do need to set up RAID at least for some stuff that I may not
>>> be able to get back.  Some videos I have are no longer available.
>>
>> RAID is not a backup solution.
>
> Not only RAID 1 isn't a back up solution, because it offers temporary
> redundancy rather than diverse protection, but under certain scenarios
you
> have a much higher chance of losing your data when the first drive
fails.  If
> you bought two (or more) drives at the same time and built a RAID from
them,
> their failure performance due to same construction and age could be quite
> similar.  On many occasions your last healthy drive fails, just as you
try to
> rebuild the RAID.
>


I think this has happened to folks on this list.  I've read about this
somewhere before.  It makes sense too.  I'd like to have two different
brands of drives if I could.  That should spread things out, maybe.

Dale

:-)  :-)


[-- Attachment #2: Type: text/html, Size: 1596 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1006 bytes --]

On Mon, 04 May 2015 03:23:48 -0500, Dale wrote:

> >> What
> >> I wish, I had a second puter in a outbuilding that I could copy to
> >> over ethernet or something.  May help in the event of a house fire
> >> etc.  
> >
> > You have, it's called Amazon S3 :) It's a lot cheaper than a second
> > computer, and a lot more reliable.

> My internet is way to slow for that.  It would take weeks maybe a month
> to upload all this stuff.  I have DSL but it is the basic package.  If I
> were on cable or had a real fast DSL, maybe.  Thing is, I really don't
> want some of my stuff on the internet anyway.  ;-)

You only need to upload it once, so it doesn't really matter how long it
takes. After that you do incremental backups. I use app-backup/duplicity
which not only takes care of incremental backups and communicating with
S3, but also encrypts everything with GPG. No one would know you were
uploading goat porn :)


-- 
Neil Bothwick

When there's a will, I want to be in it.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Dale]

Neil Bothwick wrote:
> On Mon, 04 May 2015 03:23:48 -0500, Dale wrote:
>
>>>> What
>>>> I wish, I had a second puter in a outbuilding that I could copy to
>>>> over ethernet or something.  May help in the event of a house fire
>>>> etc.  
>>> You have, it's called Amazon S3 :) It's a lot cheaper than a second
>>> computer, and a lot more reliable.
>> My internet is way to slow for that.  It would take weeks maybe a month
>> to upload all this stuff.  I have DSL but it is the basic package.  If I
>> were on cable or had a real fast DSL, maybe.  Thing is, I really don't
>> want some of my stuff on the internet anyway.  ;-)
> You only need to upload it once, so it doesn't really matter how long it
> takes. After that you do incremental backups. I use app-backup/duplicity
> which not only takes care of incremental backups and communicating with
> S3, but also encrypts everything with GPG. No one would know you were
> uploading goat porn :)
>
>


It may be only once but it would be a very large once plus I'm on my
puter a lot.  Uploading slows my surfing to almost a dead stop.  Newegg
is a nightmare for me to surf on.  Slowest thing I ever seen. Newegg
isn't alone tho.

Dale

:-)  :-) 

Re: [gentoo-user] Hard drive storage questions

[Alan Mackenzie]

Hello, Dale.

On Mon, May 04, 2015 at 03:23:48AM -0500, Dale wrote:
> Neil Bothwick wrote:

> >> What I wish, I had a second puter in a outbuilding that I could copy
> >> to over ethernet or something.  May help in the event of a house
> >> fire etc.

> > You have, it's called Amazon S3 :) It's a lot cheaper than a second
> > computer, and a lot more reliable.

> My internet is way to slow for that.  It would take weeks maybe a month
> to upload all this stuff.  I have DSL but it is the basic package.  If I
> were on cable or had a real fast DSL, maybe.  Thing is, I really don't
> want some of my stuff on the internet anyway.  ;-)

For the stuff you don't want on the internet, encrypt it!  I've recently
started using ccrypt.  It takes MUCH less time to encrypt things than it
does to transmit them over the net to a server - for my ~4.6 Gb backup,
it takes about 3 minutes to encrypt.  Sending it to my backup server then
takes the best par of an hour (at 10 Mbit/s upload speed).

I suspect your upload speed is way less, but if you had a few hundred
megabytes of really special stuff, this route might be useful.

> I'll come up with something tho.

> Dale

> :-)  :-)

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 990 bytes --]

On Mon, 04 May 2015 05:40:25 -0500, Dale wrote:

> > You only need to upload it once, so it doesn't really matter how long
> > it takes. After that you do incremental backups. I use
> > app-backup/duplicity which not only takes care of incremental backups
> > and communicating with S3, but also encrypts everything with GPG. No
> > one would know you were uploading goat porn :)

> It may be only once but it would be a very large once plus I'm on my
> puter a lot. 

You have to sleep some time, your computer doesn't :)

> Uploading slows my surfing to almost a dead stop.  Newegg
> is a nightmare for me to surf on.  Slowest thing I ever seen. Newegg
> isn't alone tho.

As long as you restrict the upload speed to around 80-80% of your
available upstream bandwidth, it shouldn't affect downloading
significantly. It's when you saturate the upstream that your downloads
are affected.


-- 
Neil Bothwick

Computer apathy error: don't bother striking any key.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Mon, May 4, 2015 at 6:31 AM, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Mon, 04 May 2015 03:23:48 -0500, Dale wrote:
>
>> >> What
>> >> I wish, I had a second puter in a outbuilding that I could copy to
>> >> over ethernet or something.  May help in the event of a house fire
>> >> etc.
>> >
>> > You have, it's called Amazon S3 :) It's a lot cheaper than a second
>> > computer, and a lot more reliable.
>
>> My internet is way to slow for that.  It would take weeks maybe a month
>> to upload all this stuff.  I have DSL but it is the basic package.  If I
>> were on cable or had a real fast DSL, maybe.  Thing is, I really don't
>> want some of my stuff on the internet anyway.  ;-)
>
> You only need to upload it once, so it doesn't really matter how long it
> takes. After that you do incremental backups. I use app-backup/duplicity
> which not only takes care of incremental backups and communicating with
> S3, but also encrypts everything with GPG. No one would know you were
> uploading goat porn :)

I tend to use a few strategies.

Typical stuff in /home, /etc: duplicity daily backups to S3.  It is
small, and safe.  Oh, and it is all on RAID too, which reduces the
risk of needing to actually restore it (RAID is primarily about
downtime, not backup).  Encryption keys are burned to multiple CDs and
stored in multiple safe places.

Photos and other valuable media:  Also gets the duplicity S3
treatment, but after every few GB I do a one-time upload to Glacier
and then remove it from my daily backups.  This stuff is write-once,
so backing it up daily is overkill.  When S3 was more expensive I
would burn two copies to DVD and store offsite, but that became a PITA
and Amazon is a lot cheaper now.  If I ever need to restore it it is
unlikely I'd need it all at once, so I can do so slowly and not get
killed by fees.

MythTV recordings, random video from internet, etc:  btrfs raid plus a
second daily rsync to ext4 (still local).  The rsync is only because
I'm still in playing-around mode with btrfs.  Once I trust it fully
I'll drop it and just rely on the RAID.  I'd be annoyed if I lost all
this stuff, but only for a week or two.  Trying to properly back up
multiple TB of media is just way too expensive and this stuff just
isn't valuable enough to care about.

I structure my filesystem around my backup strategy.  All the stuff I
really care about is in /home.  Stuff I don't care so much about goes
outside of /home and is symlinked back in where necessary.  So, I
don't need to play around with too many exclusion rules.

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Nuno Magalhães]

Greetings gents.

I may have missed it, but i haven't seen this suggested yet: RAID+LVM.
If you already have a 3TB drive, buy another (or two more) and build a
RAID1 or 5 array on them. Then build your LVM on top of /dev/md0 (or
whatever device your raid is).

Another approach is ZFS with RAID-Z or similar. I don't know how/if
ZFS splits data among the drives, but i assume it's wise enough to do
so in a way similar to a RAID+LVM combo.

If going RAID, make sure the rpm and cache are the same for
performance's sake, and you can mix and match drives from different
vendors (perhaps you should, to add to the redundancy).

I don't know about btrfs, seems like it's still in a testing-phase so
i'm not touching it yet.

Just my 2¢

Cheers,
Nuno

Re: [gentoo-user] Hard drive storage questions

[Walter Dnes]

On Mon, May 04, 2015 at 02:23:55AM -0500, Dale wrote
> Dale wrote:
> 
> Well, I read replies a few times and I think it is best to just add a
> new drive.  Heck, I've already had a 3TB drive to fail.  Anyway, I also
> need to look into some sort of backup system.  I used to do this with
> DVDs but with this much "stuff", that just isn't a good idea, not to
> mention that DVDs have their own issues.   I may take a peek into a RAID
> setup since really, that is about the best if not only way to do it. 

  How often do you need to refresh your backups?  And how much does a
medium-size safety-deposit box cost in your area?  Would the bank object
to you going into your safety-deposit box once a month?  Here's a plan...

1 computer with a main drive, and 2 backup drives.  The backup drives
are either removable internals, or standalone externals.  In either
case, they would have to fit inside the safety deposit box.

  Month #

1) make duplicate backups of your machine to both backup drives, and
stick 1 into the bank safety-deposit box, and keep the other at home

2) - update your home backup
   - take it to the bank
   - swap it with the other drive
   - bring the other drive home and update that backup immediately

3) all succeeding months... GOTO 2 (rinse; lather; repeat)

  No worry about uploading terabytes of data over a slow ADSL link.
This is basically a reprise of Andrew Tanenbaum's quote...

> Never underestimate the bandwidth of a station wagon full of tapes
> hurtling down the highway.

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-user] Hard drive storage questions

[Alan McKinnon]

On 04/05/2015 20:42, Nuno Magalhães wrote:
> Greetings gents.
> 
> I may have missed it, but i haven't seen this suggested yet: RAID+LVM.
> If you already have a 3TB drive, buy another (or two more) and build a
> RAID1 or 5 array on them. Then build your LVM on top of /dev/md0 (or
> whatever device your raid is).
> 
> Another approach is ZFS with RAID-Z or similar. I don't know how/if
> ZFS splits data among the drives, but i assume it's wise enough to do
> so in a way similar to a RAID+LVM combo.

That's a good analogy. ZFS spreads it's data out across multiple drives
using an n-drives for data plus m-drives for parity type setup. And the
whole lot forms a storage pool.

The main difference is that ZFS does all it's checksumming in software,
and it's one monolithic system. You can think of it being sort of a
combination of RAID and LVM features, but it's not implemented that way.

> If going RAID, make sure the rpm and cache are the same for
> performance's sake, and you can mix and match drives from different
> vendors (perhaps you should, to add to the redundancy).
> 
> I don't know about btrfs, seems like it's still in a testing-phase so
> i'm not touching it yet.
> 
> Just my 2¢
> 
> Cheers,
> Nuno
> 


-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Mon, May 4, 2015 at 2:42 PM, Nuno Magalhães <nunomagalhaes@eu.ipp.pt> wrote:
>
> Another approach is ZFS with RAID-Z or similar. I don't know how/if
> ZFS splits data among the drives, but i assume it's wise enough to do
> so in a way similar to a RAID+LVM combo.
>...
>
> I don't know about btrfs, seems like it's still in a testing-phase so
> i'm not touching it yet.

My understanding is that both zfs and btrfs on linux are fairly
experimental.  The codebase for zfs is much more mature in general,
though its integration on Linux is recent.  The codebase for btrfs
changes rapidly, with quite a few regressions.  I've never
irrecoverably lost data on btrfs, but it wouldn't be my first choice
for a production environment unless I basically did my own QC on the
kernel.  However, all my important data is on btrfs nonetheless (with
a full backup to ext4 daily right now).

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1273 bytes --]

On Tue, 5 May 2015 06:56:20 -0400, Rich Freeman wrote:

> > I don't know about btrfs, seems like it's still in a testing-phase so
> > i'm not touching it yet.  
> 
> My understanding is that both zfs and btrfs on linux are fairly
> experimental.  The codebase for zfs is much more mature in general,
> though its integration on Linux is recent.

It's also based on an older version of ZFS, so we can expect stability to
improve where necessary, but little in the way of new features
(unless that has changed since I last used it and Sun have open sourced a
later release).

> The codebase for btrfs
> changes rapidly, with quite a few regressions.  I've never
> irrecoverably lost data on btrfs, but it wouldn't be my first choice
> for a production environment unless I basically did my own QC on the
> kernel.  However, all my important data is on btrfs nonetheless (with
> a full backup to ext4 daily right now).

I have a similar approach, although with duplicity backups to a file
server. I have had a couple of problems with btrfs on my laptop,
connected with unclean shutdowns. I didn't lose any data but the repair
process took a *long* time.


-- 
Neil Bothwick

The considered application of terror is also a form of communication.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 1751 bytes --]

On Tuesday 05 May 2015 12:33:38 Neil Bothwick wrote:
> On Tue, 5 May 2015 06:56:20 -0400, Rich Freeman wrote:
> > > I don't know about btrfs, seems like it's still in a testing-phase so
> > > i'm not touching it yet.
> > 
> > My understanding is that both zfs and btrfs on linux are fairly
> > experimental.  The codebase for zfs is much more mature in general,
> > though its integration on Linux is recent.
> 
> It's also based on an older version of ZFS, so we can expect stability to
> improve where necessary, but little in the way of new features
> (unless that has changed since I last used it and Sun have open sourced a
> later release).
> 
> > The codebase for btrfs
> > changes rapidly, with quite a few regressions.  I've never
> > irrecoverably lost data on btrfs, but it wouldn't be my first choice
> > for a production environment unless I basically did my own QC on the
> > kernel.  However, all my important data is on btrfs nonetheless (with
> > a full backup to ext4 daily right now).
> 
> I have a similar approach, although with duplicity backups to a file
> server. I have had a couple of problems with btrfs on my laptop,
> connected with unclean shutdowns. I didn't lose any data but the repair
> process took a *long* time.

During a backup of a home directory I noticed loads of Chromium and Firefox 
crash/recovery files being copied over.  However, I don't know if this is a 
btrfs problem, or the fact that I had to forcefully shut down KDE once or 
twice recently, because the desktop would not logout/shutdown normally.

The fsck which ran when the machine rebooted did not revealed any problems.  
Is there a different recommended way for checking for fs errors?

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 753 bytes --]

On Tue, 5 May 2015 13:05:55 +0100, Mick wrote:

> During a backup of a home directory I noticed loads of Chromium and
> Firefox crash/recovery files being copied over.  However, I don't know
> if this is a btrfs problem, or the fact that I had to forcefully shut
> down KDE once or twice recently, because the desktop would not
> logout/shutdown normally.

Chromium saves its open tabs and reopens them after a reboot, even if it
is shut down forcibly.
 
> The fsck which ran when the machine rebooted did not revealed any
> problems. Is there a different recommended way for checking for fs
> errors?

btrfs check - it needs the filesystem to be unmounted and has a repair
option.


-- 
Neil Bothwick

Do you steal taglines too?

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 1084 bytes --]

On Tuesday 05 May 2015 13:21:47 Neil Bothwick wrote:
> On Tue, 5 May 2015 13:05:55 +0100, Mick wrote:
> > During a backup of a home directory I noticed loads of Chromium and
> > Firefox crash/recovery files being copied over.  However, I don't know
> > if this is a btrfs problem, or the fact that I had to forcefully shut
> > down KDE once or twice recently, because the desktop would not
> > logout/shutdown normally.
> 
> Chromium saves its open tabs and reopens them after a reboot, even if it
> is shut down forcibly.
> 
> > The fsck which ran when the machine rebooted did not revealed any
> > problems. Is there a different recommended way for checking for fs
> > errors?
> 
> btrfs check - it needs the filesystem to be unmounted and has a repair
> option.

Right, I didn't run the repair option, thinking that this is only needed if 
one receives error reports during a checking cycle.  Admittedly I am also 
being fearful of breaking filesystems (I was traumatised repeatedly when 
trying to repair reiser4 back when ... LOL! ).

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Tue, May 5, 2015 at 8:21 AM, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Tue, 5 May 2015 13:05:55 +0100, Mick wrote:
>
>> During a backup of a home directory I noticed loads of Chromium and
>> Firefox crash/recovery files being copied over.  However, I don't know
>> if this is a btrfs problem, or the fact that I had to forcefully shut
>> down KDE once or twice recently, because the desktop would not
>> logout/shutdown normally.
>
> Chromium saves its open tabs and reopens them after a reboot, even if it
> is shut down forcibly.
>
>> The fsck which ran when the machine rebooted did not revealed any
>> problems. Is there a different recommended way for checking for fs
>> errors?
>
> btrfs check - it needs the filesystem to be unmounted and has a repair
> option.

I don't think the chromium/firefox issues are in any way a sign of a
filesystem problem.  Application crashing and filesystem errors are
completely different matters.  If atop dumps core 14 times a day (as
it seems to love to do) btrfs just happily stores them in case I ever
want to look at them.

In general btrfs tends to do most of its fixing online.  I'd only run
btrfs check if the filesystem is unmountable.  I wouldn't trust it not
to do more harm than good.  For a very long time it didn't even exist,
and btrfs is a bit different from most other filesystems in this
regard.  btrfs doesn't complain if you mount it unclean - almost all
the recovery code is in the kernel and it will generally tidy up as it
goes.  This is in contrast to many other filesystems that force you to
run fsck if they were not cleanly unmounted.

I'm not saying it is broken.  I haven't really used it much.  However,
for the most part btrfs was designed around doing most of its
operations online and these are probably the more mature code paths.

That said, btrfs check without the --repair option should be
read-only, so you can always try it.  However, I wouldn't be surprised
at all if there are no problems with your filesystem (assuming you run
it after a clean shutdown).  If there were any problems, btrfs should
have cleaned them up on your last mount.  btrfs does not overwrite
data in-place in any case, so it is a bit like ext4 with data=journal
in normal operation.  And that is what I like about btrfs (and the
same applies to zfs) - the basic design of the filesystem tends to
prioritize data integrity, and thus even with all my panics and
mounting problems with btrfs, I've always been able to recover, and at
every point I could at least mount the filesystems read-only and read
everything off of them.  And a successful read on btrfs/zfs means that
the checksum matched, so the risk of corruption is fairly low.

-- 
Rich

Re: [gentoo-user] Hard drive storage questions

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 484 bytes --]

On Tue, 5 May 2015 08:53:29 -0400, Rich Freeman wrote:

> In general btrfs tends to do most of its fixing online.  I'd only run
> btrfs check if the filesystem is unmountable.

That's the only time I've had to use it. This was on a laptop with a
single SSD, so there was no where to sync good data from. It worked
without data loss but was very slow, around 12 hours for a 250GB
filesystem on one occasion.


-- 
Neil Bothwick

Top Oxymorons Number 30: Business ethics

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Hard drive storage questions

[Bill Kenworthy]

On 06/05/15 05:50, Neil Bothwick wrote:
> On Tue, 5 May 2015 08:53:29 -0400, Rich Freeman wrote:
> 
>> In general btrfs tends to do most of its fixing online.  I'd only run
>> btrfs check if the filesystem is unmountable.
> 
> That's the only time I've had to use it. This was on a laptop with a
> single SSD, so there was no where to sync good data from. It worked
> without data loss but was very slow, around 12 hours for a 250GB
> filesystem on one occasion.
> 
> 

I have two btrfs raid1 multidisk arrays (3x and 5x 2T drives) that take
"forever" to fsck and  after a power crash (had two in 2 days :( appear
to hang at the fsck stage on reboot (no disk activity and some hours
later still none.)  Ive taken to booting from a usb key and editing
fstab to get a full start and manually fsck - awkward.

Its been quite awhile now since I've found problems either with an fsck
or a scrub after abusing the arrays so I would like to force a degraded
mount on reboot after a crash. I am using genkernel and openrc - where
can I to modify the boot scripts? (looks like
/usr/share/genkernel/defaults/linuxrc but the actual place to add the
btrfs options isn't obvious)

BillK

Re: [gentoo-user] Hard drive storage questions

[Bill Kenworthy]

On 06/05/15 06:21, Bill Kenworthy wrote:
> On 06/05/15 05:50, Neil Bothwick wrote:
>> On Tue, 5 May 2015 08:53:29 -0400, Rich Freeman wrote:
>>
>>> In general btrfs tends to do most of its fixing online.  I'd only run
>>> btrfs check if the filesystem is unmountable.
>>
>> That's the only time I've had to use it. This was on a laptop with a
>> single SSD, so there was no where to sync good data from. It worked
>> without data loss but was very slow, around 12 hours for a 250GB
>> filesystem on one occasion.
>>
>>
> 
> I have two btrfs raid1 multidisk arrays (3x and 5x 2T drives) that take
> "forever" to fsck and  after a power crash (had two in 2 days :( appear
> to hang at the fsck stage on reboot (no disk activity and some hours
> later still none.)  Ive taken to booting from a usb key and editing
> fstab to get a full start and manually fsck - awkward.
> 
> Its been quite awhile now since I've found problems either with an fsck
> or a scrub after abusing the arrays so I would like to force a degraded
> mount on reboot after a crash. I am using genkernel and openrc - where
> can I to modify the boot scripts? (looks like
> /usr/share/genkernel/defaults/linuxrc but the actual place to add the
> btrfs options isn't obvious)
> 
> BillK
> 
> 

Hmm ... looks like its actually /etc/init.d/localmount ...

BillK

Re: [gentoo-user] Hard drive storage questions

[Dale]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Neil Bothwick wrote:
> On Mon, 04 May 2015 05:40:25 -0500, Dale wrote:
>
>>> You only need to upload it once, so it doesn't really matter how long
>>> it takes. After that you do incremental backups. I use
>>> app-backup/duplicity which not only takes care of incremental backups
>>> and communicating with S3, but also encrypts everything with GPG. No
>>> one would know you were uploading goat porn :)
>
>> It may be only once but it would be a very large once plus I'm on my
>> puter a lot.
>
> You have to sleep some time, your computer doesn't :)

A lot of the time, I'm downloading a list of movies while I am
sleeping.  That's when I do most of my downloading.  I use download
helper.  Sometimes it can download for several hours.  There are times
when I nap and when I wake up, it is still downloading.  ;-)  I do the
same when I leave to go to town too.


>
>
>> Uploading slows my surfing to almost a dead stop.  Newegg
>> is a nightmare for me to surf on.  Slowest thing I ever seen. Newegg
>> isn't alone tho.
>
> As long as you restrict the upload speed to around 80-80% of your
> available upstream bandwidth, it shouldn't affect downloading
> significantly. It's when you saturate the upstream that your downloads
> are affected.
>
>

I don't know how to limit that.  Still, I have a really slow upload
speed.  While I wouldn't want to lose some of it, it also would be a lot
of trouble given the large volume of data.  I'd much prefer something
local and much faster.  Now for my camera pics, that could be a option. 
Much less data and lots more important too.  I'm assuming this is what
you are talking about?

https://aws.amazon.com/s3/

I'm trying to figure out just how much this would cost here.  o_O  Just
for my pics tho.

Dale

:-)  :-)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iEYEARECAAYFAlVN58wACgkQiBoxVpK2GMAqfwCeMSL9IrGPJl5gJrI4crhk0SZn
K0MAnig1nRYvR3lB9fY8b/+ou+K3NjCh
=0DMG
-----END PGP SIGNATURE-----

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Sat, May 9, 2015 at 6:56 AM, Dale <rdalek1967@gmail.com> wrote:
>
> https://aws.amazon.com/s3/
>
> I'm trying to figure out just how much this would cost here.  o_O  Just
> for my pics tho.
>

It works out to 1-3 cents/GB/month, depending on storage tier.
Glacier is cheapest and very secure (or so they claim), but you will
pay more to retrieve the data if you need it.  If you aren't using
RAID then I probably wouldn't use glacier since it is very likely that
you'll be doing retrievals on occasion.  The most expensive figure
costs you 10c/GB to retrieve, and should be secure (again, their
claims).  The in-between figure is for reduced redundancy - it also
costs 10c/GB to retrieve, but is less secure.


I typically use the mid-cost reduced-redundancy option, since this is
intended solely as a backup.  If I were archiving data and not keeping
a copy locally I would not use reduced-redundancy.  As a backup, it is
already redundant - what are the odds of my house and the Amazon
datacenter having a disaster on the same day?  Otherwise, if their
datacenter burns down and the data disappears, then on the next day
duplicity will simply do another full backup and I'm protected again.

One thing you can't cheaply do with Amazon is verify your backups.
Duplicity will happily check the data files against the manifest
hashes with a simple command, but it will cost you 10c/GB for whatever
you verify, since it will need to be transferred out.  I guess another
option is to launch an EC2 instance with duplicity on it and have it
do the verify.  That would be an internal Amazon transfer which is
both free and much faster, but it will cost you a few cents per hour
for the CPU time.  I also don't know if duplicity can verify a backup
without the encryption keys - if it can't then you'll have to upload
your keys to EC2 which means Amazon could read your backups if they
wanted to.  Otherwise duplicity is encrypting locally and all Amazon
does is store a bunch of encrypted data and regurgitate it on demand.

--
Rich

Re: [gentoo-user] Hard drive storage questions

[Todd Goodman]

* Rich Freeman <rich0@gentoo.org> [150509 09:00]:
[..SNIP..]
> One thing you can't cheaply do with Amazon is verify your backups.
> Duplicity will happily check the data files against the manifest
> hashes with a simple command, but it will cost you 10c/GB for whatever
> you verify, since it will need to be transferred out.  I guess another
> option is to launch an EC2 instance with duplicity on it and have it
> do the verify.  That would be an internal Amazon transfer which is
> both free and much faster, but it will cost you a few cents per hour
> for the CPU time.  I also don't know if duplicity can verify a backup
> without the encryption keys - if it can't then you'll have to upload
> your keys to EC2 which means Amazon could read your backups if they
> wanted to.  Otherwise duplicity is encrypting locally and all Amazon
> does is store a bunch of encrypted data and regurgitate it on demand.
> 
> --
> Rich

Thanks for the great post Rich.

As for keys, you could use Amazon's AWS Key Management Service.
Of course they could be sitting there gathering keys, but at some point
you either have to trust they'll do what they say or simply decide not
to use them at all (IMNHO.)

You could also use AWS Key Management for backup data you want
"reasonably" secured and then your own keys for data you want more
highly secured (hopefully much smaller so the verify costs are more
reasonable.)

Todd

Re: [gentoo-user] Hard drive storage questions

[Rich Freeman]

On Sat, May 9, 2015 at 10:46 AM, Todd Goodman <tsg@bonedaddy.net> wrote:
>
> As for keys, you could use Amazon's AWS Key Management Service.
> Of course they could be sitting there gathering keys, but at some point
> you either have to trust they'll do what they say or simply decide not
> to use them at all (IMNHO.)

That is really intended more for credentials used for hosted systems
to communicate with other services/each other/etc.  If you have to
have your credentials in the cloud, then you might as well have a
somewhat secure way to manage them.  However, that is clearly inferior
to not putting credentials in the cloud in the first place.

>
> You could also use AWS Key Management for backup data you want
> "reasonably" secured and then your own keys for data you want more
> highly secured (hopefully much smaller so the verify costs are more
> reasonable.)
>

I just don't frequently verify my backups.  I'm willing to trust
Amazon to have my data when I ask for it.  That is their entire
business model with S3 and they're probably one of the stronger links
in the data security chain.  If I'm going to be paranoid about that,
I'm going to probably have other things I'd prefer to improve first.

I keep copies of my backup keys in a few places.  My thread model is
somebody hacking my account looking for personal data
(finances/keys/whatever).  If they hack into Amazon they won't have
the necessary keys.  If somebody manages to steal one of my keys in
safekeeping elsewhere, they won't have access to any of the data
encrypted using the key.  If the NSA or whoever is going to access my
Amazon data and also ask my bank to open my safe deposit box or
whatever, then more power to them.  I run a tor node, so they've
probably rooted my box anyway.


-- 
Rich