What do you do upon first getting a new server?

plumberg

As title says.. what do you usually run on the server as soon as its deployed?

I run yum update -y.. and then practically keep it idling...

Just curious to learn more about best practices...

if you can share your scripts or code it will be awesome!

Thanks in advance.

zane_

Mostly something from here https://github.com/awesome-selfhosted/awesome-selfhosted/

Or some type of Anycasting/CDN stuff.

403_Forbidden

Usually, first reinstall the operating system, Debian or Arch Linux, depending on the memory situation. Then change the SSH port and change to the key login method.

plumberg

@403_Forbidden said:
Usually, first reinstall the operating system, Debian or Arch Linux, depending on the memory situation. Then change the SSH port and change to the key login method.

Do you do custom iso install? Any specific file system or partition layout you use?

403_Forbidden

@plumberg said:

@403_Forbidden said:
Usually, first reinstall the operating system, Debian or Arch Linux, depending on the memory situation. Then change the SSH port and change to the key login method.

Do you do custom iso install? Any specific file system or partition layout you use?

Yes, I use a custom installation, usually I will separate the boot partition (between 512MiB and 1024MiB), the rest of the space is used as the root partition, and all use the XFS file system, if the VPS memory is small (below 512MiB) I will also separate the SWAP area and set the swappiness value to 1.

m4nu

Run my Ansible playbook to make myself at home. 🛀

corbpie

Turn it off and log out of the client area.

swat4

pray to god that it won't deadpool

DataRecovery

@plumberg said: I run yum update -y .. and then practically keep it idling...
Just curious to learn more about best practices...

Just get rid of "practically" and your approach will reach the perfection.

plumberg

@DataRecovery said:

@plumberg said: I run yum update -y .. and then practically keep it idling...
Just curious to learn more about best practices...

Just get rid of "practically" and your approach will reach the perfection.

Of course

tetech

Repartition and hot-load Alpine

MikeIn

Yum or apt update & upgrade
Change ssh port
Restart

sandanista

Change root port, remove password authentication and root user, enable su root user && SSH keys. Then yum update.

DP

I log in first.

plumberg

@tetech said:
Repartition and hot-load Alpine

Curious why hotload alpine?

plumberg

@sandanista said:
Change root port, remove password authentication and root user, enable su root user && SSH keys. Then yum update.

Nice. Any script that does this all?

plumberg

@thedp said:
I log in first.

Ahha. Cannot do anything unless we log in...

TimboJones

Get IP and configure DNS A name since it needs propagation time.

Also, I really don't understand why more panels don't have one time script options to configure the server right from the start. It actually makes Cloud at Cost more usable than other providers if you need to reinstall lots.

tetech

@plumberg said:

@tetech said:
Repartition and hot-load Alpine

Curious why hotload alpine?

Because there's rarely a pre-built image/template and not every provider offers custom ISO. Process works like:

Start with CentOS image -> svn co repartition script -> hot resize of partitions -> hot-load Alpine (2x) -> run setup script.

Setup script handles disk encryption, joining private cloud, and container system.

Mytonn

.

valk

Erm, configure a MOTD and run my ansible , set up an A record for the server on my domain then uhhhh idk I'll idling on it somehow, nothing to do more I guess?

chihcherng

Stop and remove unneeded network services like HTTP and SMTP.
Change timezone to UTC.
Make rsyslog use high precision timestamps.

plumberg

@tetech said:

@plumberg said:

@tetech said:
Repartition and hot-load Alpine

Curious why hotload alpine?

Because there's rarely a pre-built image/template and not every provider offers custom ISO. Process works like:

Start with CentOS image -> svn co repartition script -> hot resize of partitions -> hot-load Alpine (2x) -> run setup script.

Setup script handles disk encryption, joining private cloud, and container system.

this is interesting, thanks for sharing

plumberg

@chihcherng said:
Stop and remove unneeded network services like HTTP and SMTP.
Change timezone to UTC.
Make rsyslog use high precision timestamps.

Whats the benefit of using high precision timestamps for rsyslog?

TomBG

I'll be short.

Login, secure, bench, let it idle for 2 years.

If it is still performing well in 2 years - start using, cancel otherwise.

SteveMC

What do you do upon first getting a new server?

• Verify that the CPU, RAM and Disk are those advertised (or higher )
• Test the CPU.
• Test the RAM.
• Check the disk(s) health.
• Write the whole sectors/cells of the disk(s) one time.
• Check the disk(s) health again.
• Then reinstall the system (Debian), with only what I need, and my SSH key, and new SSH port.
• Reboot
• Check the disk(s) health, one last time.

LTniger

Inhale and hold your breath, than firewall and wireguard, exhale and setup SSH key authentication.

mrsky

@SteveMC said:

What do you do upon first getting a new server?

• Verify that the CPU, RAM and Disk are those advertised (or higher )
• Test the CPU.
• Test the RAM.
• Check the disk(s) health.

Can you explain how you do it? 🤗🤗

DataRecovery

@SteveMC said:

• Check the disk(s) health.
• Write the whole sectors/cells of the disk(s) one time.
• Check the disk(s) health again.

Proper DIY drive health test indeed involves a full write test.

But to do such test properly you need to have the target drive connected as a secondary one in a working machine. Hence you need to have two drives in the system.

dodheimsgard

I change root password to root or password and wait for abuse emails.

PUSHR_Victor

• Curse unattended upgrades if enabled
• If installing OS myself, curse IPMI as it always has some problem. Throw a tantrum at how no one can do IPMI right obviously. Curse Java and everyone who uses it.
• Add to monitoring and DNS
• Start auto configuration scripts and open a beer.
• Curse auto configuration for no real reason but being slow sometimes.
• Check that SSH keys work properly
• Confirm proper operation
• Start cache fill procedures, monitor them, move on.