Installing AlmaLinux 9.1 With Low RAM and More Useful Install Info

Jab

Hey I don't want to sounds like that guy... but many of those posts above clearly need a dedicated topic - there is some nice knowledge there that shouldn't be buried deep deep in VirMach bus spam :-)

// Flagged for mods.

localhost

@Jab said:
Hey I don't want to sounds like that guy... but many of those posts above clearly need a dedicated topic - there is some nice knowledge there that shouldn't be buried deep deep in VirMach bus spam :-)

// Flagged for mods.

I did not expect so much run around for a trivial thing. I am all in for mods to clean up and bucket the related comments in a new thread. Thanks

localhost

@dgc1980 said:
if you just want to clone a vps from one to another,

boot both systems into rescue

on the one already installed

dd if=/dev/vda bs=4M | ssh root@SECOND dd of=/dev/vda status=progress

then reboot the destination, login via ssh, I would also suggest resetting the ssh fingerprints

Yes, this worked and the system was cloned, without issue. I was able to get the LUKS disks as is. Now to figure out how to extend the partition size...

So, whats the difference between this and what you were suggesting earlier?

I would suggest using qemu locally, install it to a raw image, then ssh the raw image to the server and resize it.

Thanks

dgc1980

@localhost said: So, whats the difference between this and what you were suggesting earlier?

basically using qemu locally etc, you are just writing a image of the drive to transfer, rather than using the drive attached.

for resizing your partition, use parted print to fix the partition table, then parted /dev/vda resizepart THEPARTYOUWANT 100% then resize2fs /dev/vdaPART

oh since you using LUKS it will be different.

cryptsetup luksOpen /dev/vda3 crypt-volume
parted /dev/vda print
parted /dev/vda resizepart 3 100%
cryptsetup resize crypt-volume
vgchange -a y YOURVG
pvresize /dev/mapper/crypt-volume
lvresize -l+100%FREE /dev/VG/LV
e2fsck -f /dev/VG/LV
resize2fs /dev/VG/LV

for example

localhost

@dgc1980 said:

@localhost said: So, whats the difference between this and what you were suggesting earlier?

basically using qemu locally etc, you are just writing a image of the drive to transfer, rather than using the drive attached.

Does this have any benefit over what I did? or what are some drawbacks of what I did?

Ideally, this whole thing would not be in q if i was able to get the official ISO loaded and complete the install.

Ill try the resizing in a few. Thanks

dgc1980

@localhost said: Does this have any benefit over what I did? or what are some drawbacks of what I did?

both ways end in the same result,

ssh fingerprint are suggested to be remade

rm /etc/ssh/ssh_host_*

reboot the system and it should regenerate new fingerprints

then give the new system a new hostname and you are all good

localhost

@dgc1980 said:

@localhost said: Does this have any benefit over what I did? or what are some drawbacks of what I did?

both ways end in the same result,

ssh fingerprint are suggested to be remade

rm /etc/ssh/ssh_host_*

reboot the system and it should regenerate new fingerprints

then give the new system a new hostname and you are all good

Yup, so far, that has worked for me, things look good, but, I am now thinking, what if something went wrong during the transfer? Maybe some bits flipped?

erk

@localhost said:

Yup, so far, that has worked for me, things look good, but, I am now thinking, what if something went wrong during the transfer? Maybe some bits flipped?

There's three classes of errors to think about:

• File system/metadata - You can check for errors with something like 'e2fsck -nf /dev/mapper/vda2_crypt' - but if 'resize2fs' works, you are in good shape with what's important and usually it will get fixed on reboot anyway. That you booted means that nothing is serious wrong here

• Configuration files - again, if you are using the system, you're good here

• Corruption in the binaries - since you are up-and-running, if any binary gives an error, just use yum/dnf to remove it and redownload it then .

lemoncube

@localhost said:

@lemoncube said:

@localhost said: How do you do this in Debian?

I just checked with almalinux, and the option is there, too.

just upload your kickstart file to an easy-to-type url and submit it to this option in netboot.xyz after selecting your desired version, and hopefully, it will take care of everything else.

So, i got a kickstart config created (from fresh install) and put it on a temp web server and set the URL in netboot.xyz kickstart file location.

After that, what do I need to do? I selected the Graphical Install, and it failed.

What am I missing? Thanks

Could you try text based install instead?
(I should have done a test run myself if I had enough time)

localhost

@lemoncube said:

@localhost said:

@lemoncube said:

@localhost said: How do you do this in Debian?

I just checked with almalinux, and the option is there, too.

just upload your kickstart file to an easy-to-type url and submit it to this option in netboot.xyz after selecting your desired version, and hopefully, it will take care of everything else.

So, i got a kickstart config created (from fresh install) and put it on a temp web server and set the URL in netboot.xyz kickstart file location.

After that, what do I need to do? I selected the Graphical Install, and it failed.

What am I missing? Thanks

Could you try text based install instead?
(I should have done a test run myself if I had enough time)

It failed for text too. But again what is the correct sequence? Enter the kickstart url and then select a method of install?

lemoncube

@localhost said: It failed for text too. But again what is the correct sequence? Enter the kickstart url and then select a method of install?

I've just tested and look like it just won't work on boxes with smaller ram. The base install.img is like 700MiB which dracut will always fail to work with.
And the ks url seems doesn't work at all even when I tested on a vm with 4GiB ram. Not sure what could be the problem.
Sorry I have given you such unreliable info.

dgc1980

@Frankz, Thanks for the split, hopefully someone will find the information within here useful.

angstrom

@localhost

Just my perspective, but to me, it sounds like too much effort

My experience is that in practice, a 1GB VPS is too tight for installing AlmaLinux-9.x and family from an ISO

What would work is installing from a template, assuming that the provider makes one available

Or consider getting at least a 2GB VPS

localhost

@angstrom said:
@localhost

Just my perspective, but to me, it sounds like too much effort

My experience is that in practice, a 1GB VPS is too tight for installing AlmaLinux-9.x and family from an ISO

If the iso is mounted the install happens smoothly.

What would work is installing from a template, assuming that the provider makes one available

Haha, with @VirMach it's a dream come true. One day.

Or consider getting at least a 2GB VPS

Next buy will always focus on this number or more.

But for now, I have found it pretty easy, where I just dd the image from a freshly installed system and so far it seems to have worked... albeit I would have hoped it would be more cleaner via the iso.

Thanks