CentOS 7 kernel upgrade post EOL

Posted by nomadewolf@reddit | linuxadmin | View on Reddit | 29 comments

I know i was dumb to let it come to this point, but here we are...

My personal server has CentOS 7 installed and i'm trying to migrate it to a newer version.

In order to do so, i want to backup my data to an external USB drive.

The problem i'm facing is that, since we're talking about 5TB of data, it's taking ages to do so, sometines at a few KB/s speed. It took over 24 hours to backup 500GB.........

I'm using rsync because i want to preserve the original timestamps.

In order to maybe speed up the process, it occurred to me to install a newer kernel.

But the repos are down and that's a no go.

Migrating to Alma or Rocky is also a no go, because i have less than 20GB of free space.

I'm looking to me fellow redditors for ideas.

Cheers!

[-]

Iseeapool@reddit

Do you know Elevate ? The run elevate and migrate to whatever RHEL based distro you want in no time.

[-]

Iseeapool@reddit

Which part ? Can you send the result for df -h ?

[-]

nomadewolf@reddit (OP)

The / and /boot part.

I have an array for /home.

[-]

Iseeapool@reddit

Ok. Can you please show a df -h result ?

[-]

Ok_Size1748@reddit

Use dstat/atop to troubleshoot your slow backup speed. When you know what is the reason, you can think about kernel switch or whatever

[-]

nomadewolf@reddit (OP)

Thanks for those commands, as i didn't know about them.
Found this site with some other insteresting ones.

5 Tools for Monitoring Disk Activity in Linux - OpsDash

atop was the most helpfull as it shows 100% disk busy (destination), could this mean the disk is bad? It's only a few months old...

[-]

Ok_Size1748@reddit

How is the disk connected? Usb 2? Sata? Is a hdd or a ssd?

[-]

nomadewolf@reddit (OP)

It's a USB 3 hdd drive. That's why i thought it might be the kernel.

[-]

ultratensai@reddit

Any errors logged in smartctl and dmesg?

[-]

draeath@reddit

I doubt this has anything to do with the kernel version.

[-]

nomadewolf@reddit (OP)

You guessed correctly. Seems to be the USB drive itself.

[-]

You need to change "/etc/yum.repos.d/CentOS-Base.repo" for the one that is EOL, to keep getting updates, check that to change the kernel, you will have to do it all manually, since the load of the new rpm, to the grub. I've tried, many versions of Linux, I tried, Rocky 9, Alma, and CentOS 9 stream, and there's no competition for CentOS 7, no matter it's out of mantainance, things keep working the same, from 10 years ago... Even they said, there's code in postfix, since 1993... things like that... as far as it works with the special standard to build it all handrolled... its amazing... cheers...

[-]

nomadewolf@reddit (OP)

I already did that. Those are the Vault repos, there's no Vault repos for kernel, unfortunately.

[-]

HLingonberry@reddit

Try switching to rclone instead of rsync, it’s multi threaded and works much better for small files. With that said, if you have 100% disk busy on the target node already it won’t make much of a difference.

[-]

nomadewolf@reddit (OP)

Thanks for the tip. I'll try and do some tests.

[-]

IridescentKoala@reddit

You don't have backups? With that slow transfer speed I would buy a server built in this decade and mount the old drive.

[-]

nomadewolf@reddit (OP)

No. Those are my backups.

I have everything on each family PC and that server acts as backup.

And it has RAID.

I should have an extra, but i can't afford it.

[-]

craigleary@reddit

Kernel is probably not the savior here. Could be slow drives or lots of small data and meta files. You can move the drive to a new system as an option. If it must be copied maybe you can make some changes to rsync: upgrading just rsync on centos7 , size only switch and using no time on your mounted drive. If it’s ext4 and lot of files make sure dir_index enabled.

[-]

nomadewolf@reddit (OP)

You're correct. From my tests it seems to be the USB drive i'm using. I'll try to format it into NTFS and do some tests in Windows to see if anything is wrong. SMART reports everything ok.

[-]

_mick_s@reddit

20gb is plenty for the upgrade, and the process doesn't touch your data anyway.

Now I do agree having a backup is very much a good idea but worst case I can think of is it doesn't boot and you have to reinstall (probably on another drive) and attach your old disk to access data.

Also alma has mirrors of CentOS 7 repos which you have to use before running elevate to get to latest CentOS 7 packages.

You can just add the repo and do only CentOS 7 upgrades.

[-]

nomadewolf@reddit (OP)

I said less than 20 because that's what's recomended, but it's actually just 8...

[-]

just_some_onlooker@reddit

Well ...Almalinux has leap?

[-]

nomadewolf@reddit (OP)

No space to upgrade.

[-]

dhsjabsbsjkans@reddit

you should be able to install a newer mainline kernel. find a repo for a newer mainline and install it. But I don't think this is going to do much for you. If anything, you mike look into something like mrsync. It's a python wrapper to parallelize multiple rsyncs.

[-]

nomadewolf@reddit (OP)

I have a brand new 512GB SSD that i bought to install the new OS, will try that.