Which installation should be good?

English Install/Boot/Login
#1

Hello,

I have recently downloaded openSUSE 13.1 KDE again (because earlier aria2c issue was there) and now wish to install KDE again (at first). I have two doubts to be asked:

[1]. Should I go with LVM? Because many say they do go with it and there must be some advantages of doing so…

Also, if I go with LVM, partitions can be made automatically or I necessarily need to allocate space for each partition?

[2]. There is an option of encrypting the whole system during installation, should be selected? If I select this, everything I would have in my PC would be encrypted, so that must be perfectly secured. However, sometimes recovery becomes typical, is it true? Or encryption can be done after the installation?

Thanks.

#2

Only use LVM when you see a clear case for it and are prepared to learn to understand it.

So you must decide what you want (glue-ing disks/partitions together, RAID of some kind, both). Then you could ask for advice (LVM, MDRAID, …).

#3

Well, I am ready to prepare for learning it because I heard its many advantages. Since I get time not on regular basis, but still I would try to manage to look at it. Also, I don’t know about RAID and why to use it. I simply have a 250 GB hard disk with 2 GB RAM and nothing more in my PC.

#4

I do not know which advanrtages you heard of. May be they are only advantages for some situations and not for your’s. It is hard to decide when you do not know yourself what is attractive.

In any case, having only one disk, I do not see much advantage in using LVM. Also using DMRAID with for e.g. RAID 1 of you data gives only limited advantage because when your disk is gone, it is gone, RAID1 or not.

Examples from my usage:

One (old system) has three small disks. I made one Logical Volume out of them and then created one large file system on it. Thus this file system spreads over all three disks (used for data, not the system). It has the advantage that it offers one big flexible amount of storage instead of three small sized file systems where you allways have the problem that one is full and the other two have space left.

On another system, I have two disks off a different size. I wanted some security on the data and decided to go for RAID1 (mirroring). I could have gone for LVM again, but this time I used the “build in” Linux DM features. I partioned the larger disk in two partitions, the smaller one for the system and the larger one, of the same size as the other disk. Then I created a RAID1 for that parition and the one partition on the other disk. Now the data is protected for a failing disk. The system must however be reinstalled if the larger disk fails.

But all these are solutions for a case one has. One does not use these solutions simply because they are available and someone says that they are nice. My opinion of course.

#5

Thanks for this clarification. Basically, as conclusion, I should use LVM if I have more than one hard disks and then it shows its advantages. Ok.

During installation, there is just an option to be ticked (or not, depends) using LVM… I just thought it could be used on the single hard disk also because of the flexibility…For example, if one plans to install another distro in the near future, its easier to shrink the space using YaST, if LVM were used. But since I am not aware of dual/tripple booting system and have to keep only one distro…, I am sure, I should not go with LVM then. This just came in mind because the option was depicted there!

#6

There could be reasons to use LVM on the one and only disk of a system. Maybe when someone comes with a special requirement that then somebody would say: “Hey, you can realize that using LVM in such and such way”. But at the moment I can not produce such a case out of my head.

Changing the sizes of Logical Volumes is IMHO not easier then repartitioning. In both cases the difficult part is the data on it. Can the file system on that container easily and in a safe manner be shrinken and/or enlarged. And even when that is implemented in a tool, that will only work when the start address of the container stays on the same place (you can only shrink/enlarge at the end of a file system). Every other reorganisation requires saving of the data before the repartitioning/LV-recreation and restoring it after.

#7

So, finally it would be best for me to not go with LVM.

#8

My personal advice would be not to use it. In your case it would only make things more complicated.

#9

I use LVM on a single disk. I decided for it, because I didn’t know where my space would go. I set up a more “fragmented” partitioning than usual: A separate boot partition outside the LVM (to prepare for full system encryption - it was never applied), a /srv partition, a /tmp partition and so on - the goal was to have a system where users could misbehave while I could be sure that no data could fill up disks and crash the system. Then I left 25% of the disk unallocated. That space can easily be added to any of the partitions under the LVM hat. This has worked fine. But to tell the truth, I never have had any use for the spare disk storage. It also seems that LVM takes a performance hit, but I have not done any specific timings of it… But see below. THAT’s where I made use of LVM…

Originally, I only enabled LVM to learn it. I got the hang of it (I think), then left it alone for a while (since 12.1 arrived). A few weeks ago (it was inside a VM), i needed to shrink the VM. Since this was LVM, I could establish another disk of the size I wanted, then transfer all data (including root, system files and all!) to the new disk, shrink the old disk and transfer the system back. Everything was done while being online! How neat is thatlol!! The only downtime needed was to reinstall Grub, as I had a separate boot-partition outside the LVM. I used caf4926’s excellent article on reinstalling Grub to do that. Shrinking the system’s disk took the better part of a day, but still had a cost of less than 30 minutes downtime!

My goal when enabling LVM, was simply to learn it, as a way to learn to set up a robust system. The “online shrinking” was just a fluke, and not strictly necessary in my environment. But it demonstrates the power.

My experience, though, is if you get problems, things like LVM and encryption often is the difference between being able to remedy it or not. So I generally don’t implement such things unless there is a specific need for it. So there’s many ways to see a robust system. One “robustness model” may exclude the other. It all depends on your needs. I have no difficulty in following hcvv’s advice here.

dayfinger

#10

I agreeTrying to understand what LVM is, one needs a sort of test environment and start reading man pages and get the feel of it. I do not have the idea that the OP here is willing to invest in that. That is not wrong at all, because a normal set-up will function fully to his expectations. When LVM would have many advantages above using partitions directly, I am sure that these forums would be full of advice to go for it. Which is not the case. It is a fine thing when one digs into the area of computer room systems (often called servers) with improved up-times, disk error rcovery and the like.

And yes, your case of having space left that where you can then cut out a new partition that you can then add to an existing LV (and then increase the file system) can be an advantage when you are not sure which of your file systems may need more space in the future.

#11

I use encryption. My system is installed in an encrypted LVM. That’s the only reason that I use an LVM.

My encrypted LVM has three volumes: root, home, swap. You can probably guess what those are for.

At install, a separate “/boot” will be needed. That is not encrypted. The kernels and “initrd” stuff go in “/boot”, and they setup the encryption for the rest of the system.

If you go with encryption, then it’s a good idea to learn how to do it manually.

For example, I can boot a live DVD. I then use “cryptsetup” to make the encrypted LVM available. And next I use “vgscan” and “vgchange” to make the LVM volume accessible. And then I can mount them. I suggest that you learn how to do that, and practice it occasionally. It’s always useful to be able to repair your system if something breaks. You might have to wait till after the install to do the practicing.

#12

On 2014-02-01 18:36, hcvv wrote:

> I do not know which advanrtages you heard of. May be they are only
> advantages for some situations and not for your’s. It is hard to decide
> when you do not know yourself what is attractive.

I stay away from LVM because I simply am not confident I can repair it
in case of disasters. As simple as that: I have not invested time in
learning it in depth, and it is delicate to recover (first track lost,
entire disk lost). Well, there are backups to cover that.

There are important advantages.

You can easily add space to partitions.

You assign some initial space, and leave ample space out. As you need
more space to a partition inside the LVM (ok, not a partition, has
another name), you simply add it from the free space pool. You need that
empty space pool for this important feature. You can even add space from
other disks (adding vulnerability, too).

(On the other hand, traditional partitions can be located in case of
disaster, by gpart. LVM spaces, can not)

Another important use case for LVM is full disk encryption. It is done
by YaST. There is a boot partition, an encrypted LVM container, and
inside it all the “partitions”. You only enter the passphrase once.

(It would be possible to encrypt the entire disk with discrete
partitions, but YaST does not support this method).

My main recommendation: unless you really understand it, do not use it.

–
Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

#13

I very emphatically agree with Henk on this point. You should not use LVM at this stage of the game.

#14

I agree that it should be used only if I learn LVM or else the normal things would also take time if anything goes wrong. Unless I can spent more time on LVM, I conclude that I would have to use the simple default installation which I did for 12.3 also.

But I agree with the users who used LVM in the sense that it is really powerful. The saying goes correct ‘unless we break things, we cannot learn’. Like dayfinger, I also wanted to learn LVM, but not now as I would try it some time in future, preferably on a spare machine.

Thanks.

#15

Well, I have finally installed opensuse with the simple configuration. Here is the output of “lsblk” command:

NAME   MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
sda      8:0    0 232.9G  0 disk 
├─sda1   8:1    0    21G  0 part /
├─sda2   8:2    0   5.9G  0 part [SWAP]
└─sda3   8:3    0   206G  0 part /home
sr0     11:0    1  1024M  0 rom

But during installation I saw a line “Setting type of partition /dev/sda2 to 82”. Does it indicate something wrong because ‘82’ is something unknown…?

Also, should I run the both the following commads on regular basis?

zypper up
zypper patch

Thanks.

#16

Nope.

82 means Linux swap / Solaris
83 means Linux

That is the partition type is set to 82 for your Swap partition and 83 for all Linux type file systems.

#17

zypper patch is the same as YaST > Software > Online update. It does apply the official updates tthat are offered through the two Update repos. Mostly Security and also Recommended for the official release.

zypper up will install all packages which have a newer version (without vendor change) in all the repos you have. That includes the zypper patch ones thus doing zypper patch right before or after a zypper up is useless.

As allways, what you do is your personal responsability as a system manager.

I would certainly advise to do zypper patch on a regular base. It keeps you secure. I (and that is of course personaly) do YaST Online Update once a week (in the maintenance slot), so I get first a quick overview about what is going to happen, but in almost all cases I then click Accept, thus a blindfolded zypper patch is not very different.

Also once a week I use YaST > Software > Software management and then the Repositories View. I then choose Packman and in the list of packages at right, I right click and go for “All in this list > Update when a newer version is available”. I also check for red lines there and try to understand why they are there (often a small glitch of the Packman packers or a fall-back to an earlier version). This is the same as zypper up from Packman only.

I do have a few other repos with only one or two packages. I have disabled them normaly. But once a while (every half year?) I enable them to see if thjere is something new. These packages are not very security prone, thus when they function, why updating?

#18

Ok thanks. They defined terminology.

#19

But if you have already run zypper patch, why do you check that from YaST again? Probably for assurance…(?)… Because as you said, the same task has already been done by the command.

But even if I run “zypper up” on regular basis, I don’t think (after knowing the difference) that there is any security risk, rather I would have the latest versions and this won’t also create any conflicts because those are downloaded from the vendors already fixed (those with the current version of those apps). But its true that if only one or two packages are there, then one might go on yearly basis but the repos like Packman provides so many apps that one may prefer monthly running the same command.

#20

On 2014-02-03 16:36, panchparmeshthi wrote:
> But even if I run “zypper up” on regular basis, I don’t think (after
> knowing the difference) that there is any security risk, rather I would
> have the latest versions and this won’t also create any conflicts
> because those are downloaded from the vendors already fixed (those with
> the current version of those apps).

Nope. With “zypper up” there is some risk of some thing that stops
working. Depends on your repo combination, bad luck, etc. With some
combinations it can be frequent, with some it is never.

Consider that people maintaining important machines never do updates on
them, and when they do, they have a backout plan, and have tested those
updates on other machines.

Example: someone told me that the NASA does absolutely no updates to any
machine used on operations for weeks before the launch. No matter if you
have problems that an update solves. No updates at all.

–
Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)