Additional Linux Commands & Guides

Additional troubleshooting commands & help guides for software installed on systems.

Small Commands, Issues And Guides - A Collection

This page is for commands, issues or guides which don't need to be added as a separate page. You can search for them with Ctrl + F or click the "Search" at the top.

Only Allow PublicKey Authentication Overriding Password Auth

No idea why, some systems are completely ignorant and ignoring the demand for no password auth.

Add this line to bottom of SSHD config. No idea why, but the The PA parameter is no and ChallengeResponseAuthentication is also no, but still auths with password... sigh.

AuthenticationMethods publickey

Give A MongoDB User Permission To Command "mongodump”

use admin
db.grantRolesToUser( "USER", [ "readWrite" , { role: "readWrite", db: "config" } ] )

Remove Apache2/Nginx Python3’s modules for Certbot

Apache2: python3-certbot-apache
Nginx: python3-certbot-nginx

Remove a package that is being stubborn and will not remove

sudo dpkg --force-all -P <package-name-to-remove>

Absolute headache with MySQL 8.0 mixing with MariaDB’s installs

MySQL 8.0 has the config file in /etc/mysql/mysql.conf.d/mysqld.cnf like MariaDB has /etc/mysql/mariadb.conf.d/50-server.cnf.

BY DEFAULT MySQL binds publicly… MariaDB binds to internal localhost.

Use mysqld --verbose --help | grep bind to actually see it bind publicly on first installation…

If you mix MariaDB’s config files + MySQL’s, it’s not my.cnf, mysql.cnf, the conf.d folder entirely or any other configuration. It is literally mysqld.cnf in the mysql.conf.d folder. What utter BS.

Fixing the repository error for changing it’s “version” for updates

Example: N: Repository 'http://deb.debian.org/debian bullseye InRelease' changed its 'Version' value from ‘11.7' to '11.8'

One command: apt-get --allow-releaseinfo-change update

Taken from: https://www.reddit.com/r/debian/comments/ca3se6/for_people_who_gets_this_error_inrelease_changed/ 

The certbot command that actually works to change the email

certbot update_account –email <email>

SlimeWorldManager Installation

Taken from: https://github.com/cijaaimee/Slime-World-Manager/blob/master/.docs/usage/install.md

Crons not running for WHMCS on cPanel - They hang and sit there doing nothing

Taken from: https://support.cpanel.net/hc/en-us/articles/360052237994-How-to-increase-the-Max-Execution-Time-directive-for-PHP

RAID Ubuntu 20.04

Taken from: https://kifarunix.com/setup-software-raid-on-ubuntu-20-04/  

Note - If you get error: partition length of [BIGGERSIZE] sectors exceeds the msdos-partition-table-imposed maximum of [SMALLERSIZE]

Follow: https://askubuntu.com/questions/84538/trouble-creating-3tb-ext4-partition-due-to-msdos-partition-table-imposed-error

Switching RAID 1 to RAID 0

Scroll down a bit to locate the answer from nmr, which includes removing the mirror, changing, waiting then resizing.

Taken From: https://serverfault.com/questions/915284/is-it-possible-to-convert-raid1-to-raid0-without-system-reinstalation

Remove .html/.php extension

Read the answer from Arnon, the big threaded part with explanations.

Taken from: https://stackoverflow.com/questions/38228393/nginx-remove-html-extension

Can't run a module or build a NodeJS app due to no such file and has no installation candidate

One simple command to clear out the old NPM packages already downloaded from cache.

npm clean-install

You can then go about reinstalling with yarn build or another method used.

Any NPM package doesn't install into /usr/local/bin, such as ghostcms - Updating the prefix location due to weird bug

Taken from: https://askubuntu.com/questions/1102579/using-npm-to-install-file-to-usr-local-bin 

For some weird reason, on some installs of NodeJS and then installing npm, it ends up at /usr and not /user/local to allow /user/local/bin packages to be viewable, such as GhostCMS's ghost-cli. Stupid bug, no idea why. Run the set prefix command to fix it, then install Ghost's CLI fine.

The reason why this needs doing is due to the error "Error: Cannot find module '/usr/local/bin/ghost'" and "MODULE_NOT_FOUND" which is absolutely idiotic since you literally just installed it, but in the wrong place. This will happen when trying to start the service and it fails.

root@S01:/var/www/website.com# npm get prefix
/usr
root@S01:/var/www/website.com# npm config set prefix /usr/local
root@S01:/var/www/website.com# npm get prefix
/usr/local
sudo npm install ghost-cli@latest -g
Date command in linux - Echo and crontab support

When using the date command in terminal, if you are echoing the line, you need to make sure you don't escape the percentages as it will include the slashes.

However, when inside crontab, you will need to escape them otherwise cron believes it is an operand and not a variable to pass. And example is below where you also need the path in a full string for it to echo properly and understand the date variable.

You can also use -d for naming how we name things, such as "yesterday" "tomorrow" or "2 months ago".

Transferring server data between two dedicated systems - Wget & Python3

The easiest way to transfer a tarball or zip file from one dedicated system to another is by using wget and a python3 webserver, protected from sending out the data from UFW allow in only that IP address or using the internal network if your system has it, such as OVH's vSwitch/vRack system.

Open up SSH. In this example, we will be using Pterodactyl's UUID's in the volumes folder, but this also applies to any zip/tarball you want to do this on such as servers stored in /home or /srv.

Go to cd /var/lib/pterodactyl/volumes and then do ls to see the containers.

image.png

We will use "031d5..." as an example. Run tar -zcvf SERVERNAME.tgz 031d5 then press tab to autofill the entire server ID.

image.png

Let it finish tarballing. You'll know when it's finished as the output stops showing you what files it's archived and goes back to the normal terminal line waiting input.

image.png

Do ip a on your destination system and find the public IP. Remember, some systems have internal networking so you might be able to do "wget 10.10.10.1:8000/hub.tgz" for example instead of the public IP and whitelisting it, as it normally is already whitelisted. 

Once you've found either the public or the private IP, go back onto your source system and type ufw allow from x.x.x.x changing the x's to your IP.

You can now type ip a on your source system to get its IP you need.

On your source system, now type python3 -m http.server to start up a temporary HTTP server on port 8000 for WGet to pull from. If the port is already in use and errors out, type the port after the command, as shown below.

image.png

Go to your destination system and go to a folder you want to import into, such as making a directory called /import then going into it. Run wget x.x.x.x:PORT/FILENAME.tgz, replacing the x's with your public or private IP, the PORT with the port it opened and the FILENAME with the tarball name.

You will see a progress bar with it transferring over. Once it's done, it will go back to the normal terminal line.

You can now go back to the source system and "Ctrl + C" to exit out of the webserver.

image.png

On your destination system, you can now do what you need to do on your tarball, such as move it to another place, extract it with tar -xvf FILENAME.tgz and more.

image.png

If the transfer was successful and the extraction also was successful, make sure you go onto the old source system and delete the old tarball, you don't want to fill up on disk space.

image.png

Cannot Access UDP Connected Games Due To OVH Firewall

If you have purchased an OVH system from one of their "GAME" series, such as the GAME-1, GAME-2 or the GAME-LE-x servers, you will need to login to their interface to disable the UDP firewall called the "Game Firewall" as this will block connections such as Minecraft Geyser.

This guide presumes that you've checked multiple port open sites and they all state closed, you checked and confirmed the port is open on the firewall internally on UFW or the IPTables rules and if running the Pterodactyl panel, you've confirmed that the port is assigned to the server itself and rebooted the container to confirm working.

Login to your OVH account and click the "Bare Metal Cloud" in the top left, click "Network" along the left hand sidebar then "IP" under the dropdown.

image.png  

In the list, find your server IP and click the three dots on the right and click "Configure the GAME firewall". If you are unsure which server it is, click the blue text of the IP and it will jump you to that server to confirm it is the one you want to configure.

All you now need to do is click "Deactivating the restriction" along the top bar. 

Wait about 2-5 minutes and it should be off. You are now able to connect to that IP with ports broadcasting UDP traffic, like Minecraft Geyser. If it still doesn't work, do it again, as in enabling and disabling it. It's super temperamental and needs it to happen more than 60% of the time!

Accessing A Website Or Control Panel Shows "Too Many Redirects" Error

If you see this error:

Go to CloudFlare and click your domain.

On the left hand side, expand SSL/TLS then click on "Overview".

Set the SSL/TLS encryption mode to "Full" instead of "Flexible".

Wait about 30 seconds for the change to propagate, then you should be able to access it again.

Error 1014 Tebex - CNAME Cross-User Banned

If you are ever presented with the following below when trying to access your Tebex store using your own custom domain, such as https://store.vanillaplus.uk, this means you need to renewal your Plus plan.

You haven't been banned or suspended by CloudFlare, even though the word "banned" appears. What this means in simple terms is when Tebex tries to redirect from your custom domain to the actual store link, they stop that redirection because you haven't renewed your "Plus" plan, so the CNAME becomes "Banned" to stop the "hop" from your custom domain to abc123abc123.tebex.io (The CNAME you made).

Hopefully this provides more context. All you need to do is go and pay. After about 5-10 minutes, the error disappears.

Why Is X Not Installed On The System - How Docker Images Work

When connecting to your systems via SSH, you may notice that certain applications don't exist, are not installed or not working as they should be. This is due to how docker works and it's image system.

If you are a client who has Pterodactyl installed, servers are managed in containers, which are managed by the software called "Docker". Each server is contained in a container, simply referred to as a "house." Those little houses live together to run anything you wish to in their own little "building". They can't talk to other "houses" (Unless internal, 172.18.0.1 or your own subnet), they can't interact with them, they can't touch them, they are separate. They can run a magnitude of software such as Java, NodeJS, Python, SteamCMD, the list goes on. Now, these are managed with docker images. Images are released by many users online to help streamline the process of running a certain application. For example, let's take a java one from Pterodactyl's repository on GHCR:

What we can see here is that it's using the website ghcr.io, which is where people can share and publish docker images. Next up is /pterodactyl, which is the author of the images, then java_17, meaning that Java 17 will be installed into that container when deployed. Deploying means that you've gone to your panel, created a server, selected Java 17 as the image, and the "house" gets "built." Using this analogy should hopefully now make sense. Nothing is installed on the root level, doesn't touch the system whatsoever and doesn't interfere with other houses.

This means that when you run java --version, node -v or other such command, it won't show anything up unless we installed that application at some point for testing or development. Any questions, ask the team.

PHP Extensions Locations For .so Files

PHP stores its .so extension files in /usr/lib/php/ then the API version of PHP. For ease of use, they have been listed here as they are not listed on any site.

If you need it and you can't find it, run php -i | grep API and you will see the API number:

image.png

Installing Modules On cPanel For WHMCS - IonCube, FileInfo & Soap

IonCube

Soap

FileInfo

Referenced from: https://support.cpanel.net/hc/en-us/articles/1500000107181-How-do-I-install-Fileinfo- 

If you wish to install extensions for a user of cPanel but don't want to install these extensions globally, follow:

Note: On all instances above, you can restart Apache's FPM in the WHM interface instead of the command by searching "Services" in the top left then finding "Apache FPM".

LiteSpeed WebServer With Pterodactyl (CyberPanel)

Setup main domain then “panel” as another subdomain. During the subdomain creation, make the location of the files /panel/public. Pull files from this directory back one into /panel and make sure Ptero's /public is the folder that CyberPanel is pointing to for it's VHost - This tricks CyberPanel into loading it.

Remember it's a laravel app, so you still need to install composer, run the composer install, sort out the database connections, fix the initial permissions from the Ptero docs, install redis and sort the pteroq service (including wings if adding it onto this system).

If you get error 500, click fix permissions in file location /panel and /panel/public in the CyberPanel files interface along the top.

Set .htaccess rules in the rewrite section of CyberPanel with:

RewriteEngine On
RewriteBase /
RewriteRule ^/index.php$ - [L]
RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d
RewriteRule . /index.php [L]

Note: If not added, the panel will just 404 as it cannot redirect .php files to friendly URL’s.

For wings, it will fail since firewalld is blocking the ports. Go and open the ports in /etc/firewalld/zones/public.xml and update the public zone. Then systemctl restart firewalld. You can also do this in the Firewall page of CyberPanel along the left hand side.

Searching Through Files - Finding A Particular String And Updating It

Main command: grep -rnw "xA" -e "xB" | xargs sed -i 's/xC/xD/'

If you see on the example below, you would use the grep command up to the pipe symbol first without -l, then add it in afterwards. This is used after searching for strings. The letter L is for grep’ing just the directory location where it finds the result and not actually showing file contents.

SED WILL NOT work if you do not add -l into the -rlnw. It has to require the absolute paths and not anything else it outputs.

An Example

Searching through LuckPerms configs in all servers to replace connection timeout string.

grep -rlnw "connection-timeout: 5000" */plugins/LuckPerms | xargs sed -i 's/connection-timeout: 5000/connection-timeout: 20000/'

Installing SQLite3 To SQL with Python3 Package To Convert SQL-Lite DB to MySQL for import

Start off with pip3 install sqlite3-to-mysql, making sure apt install python3-pip is installed. Once done, upgrade pip3 with pip3 install --upgrade pip then pip3 install packaging. You can now run sqlite3mysql --help.

For simple syntax, it is:
sqlite3mysql -f <file-of-sqllitedb.db> -d <database-to-import-to> -u <sql-user-to-import-with> -p

When I tried using -u root, it didn’t work and said access denied even with correct password. Use a SQL user which has access to that particular database or a global admin. No idea why.

Works fairly well and have had no reports of issues.

An example below:

ItemsAdder With CloudFlare Proxied Domain For IP Protection

    self-host:
      enabled: true
      server-ip: https://texturepack.domain.com
      pack-port: 20000
      append-port: false

Get the “/dev/disk/by-id” information for FSTab from the udevadm command

To start with, you will need to use the disk name only without the partition name. When you go into /dev/disk/by-id and investigate the names, it will be named “part-1, part-2” etc at the end of the ID name. Ignore these for now. Do fdisk -l and get the disk name.

Run the below command:

udevadm info -q symlink --path=/sys/block/<DRIVE-NAME> | awk '{print "/dev/" $1}'

Repeat this command until you have “nvme-eui.ID” or "sci-ID" as the first time you run it, it may list the “by-path/” statement which isn't helpful or the actual “by-id/DISKNAME” which is what we want.

Once you've found the ID, go to /dev/disk/by-id and look for the partition you want from fdisk -l. Use this to correlate the lists together. 

image.png

For example, /dev/sda3 is going to be "part-3" from the ID list.

DO NOT IGNORE THIS WARNING. IT IS CRUCIAL TO READ.

Note: Make sure you set the drive to ext4 before rebooting so it actually knows how to handle the partition. If you don't, the partition won't actually mount and when you reboot you will get put into initramfs, so have rescue mode on standby with a KVM!

8c1d2ab5-69a4-4d2d-994b-a751b425dbf2.png

Go into nano /etc/fstab and add the /dev/disk/by-id/IDNAME-PARTX into the FSTab. Example below for nvme0n1p1’s part 1, the partition for the datastore, ext4, defaults, 0, 1.

Resizing single volume group and logical volume after disk expansion

This guide is used after cloning to a increased size disk such as 500GB to 1TB, or by having some space left over from installing the OS, such as the rest of the disk but only 50GB allocated to the / partition.

Firstly run vgdisplay to get the volume group information.

image.png

Run lsblk and notice that the size of the partition on the disk and the volume group is 424G, not the 931.5G. 

Note: After resizing, it will be a bit smaller than the disk as the other partitions take it up, so around 926G at the end

Let's add the rest of the free space to the partition first. Run cfdisk /dev/nvme#n1, replacing hashtag with drive number. No need for a partition here as we need to do changes on the disk itself.

I should have an image here of CF Disk at some point when I have a free size again.

Go down to partition 3 (Or where your logical volumes are stored for expansion) and chose "Resize". Just put in the amount you wish to add to that partition (which is presumably max), then write it.

Once you've done this, run pvresize /dev/nvme#n1p#, replacing both the hashtags respectively.

Next run lvextend -l +100%FREE /dev/vg#/LOCATION, replacing the hashtag with the volume group number, normally vg0, and location with the mountpoint name, normally "root". An example:

This will say resized. I don't have an image currently of this.

image.png

Once done, run resize2fs /dev/vg#/LOCATION, replacing the values again with the relevant fields.

This will say "online sizing required" and resize.

Run df -h and lsblk to confirm resizing.

Resizing a logical group then volume within a parented RAID number

Taken from https://packetpushers.net/blog/ubuntu-extend-your-default-lvm-space/ 

Run cat /proc/mdstat - Get the RAID partition you want to update after doing all the resize commands to switch from Raid 1 to Raid 0.

Taken from https://serverfault.com/questions/915284/is-it-possible-to-convert-raid1-to-raid0-without-system-reinstalation 

image.png

Next, get the partition info which should look like this with lsblk:

image.png

Notice how vg0-root and vg0-swap are sitting on on both disks partition 3 parented under RAID mdadm md127? This is what we will be working on.

Run vgdisplay - Get Volume Group name and the path, in this case will be "/dev/vg0/xxx". Sometimes it is "/dev/mapper/ubuntu–vg-ubuntu–lv", different on how the OS was installed.

image.png

Increasing the size now comes in 3 steps

1. We now need to physically resize the volume of the mdadm RAID by doing pvresize /dev/RAID# as shown below:

image.png

2. Expand the Logical Volume by running lvextend -l +100%FREE /dev/VGNAME/LVNAME. We worked out the VGNAME was vg0 and we worked out the LVNAME is root from lsblk - this showed us vg0-root. It should be successful.

image.png

3. Your final stage is doing the typical resize2fs /dev/VGNAME/LVNAME. You do NOT do it on the mdadm RAID#, as this will just fail with not finding the superblock:

image.pngimage.png

Check with df -h, and you will see it has been resized successfully:

image.png

CloudFlare does not minify javascript correctly when working with Grafana

671244d4-7564-41d2-875d-503ac657a372.png

Speed > Optimisation > Content Optimisation.

Follow the below if you want to disable JavaScript Minify'ing for just that website, not the whole domain:

Install Custom Version Of MySQL Server

In this guide, replace .27 with what version you need. At time of writing, Version .37 is now out, but on and after .29 breaks games such as Unturned with their charsets.

To begin, download the bundle.

wget https://dev.mysql.com/get/Downloads/MySQL-8.0/mysql-server_8.0.27-1ubuntu20.04_amd64.deb-bundle.tar

Extract it.

tar -xf mysql-server_8.0.27-1ubuntu20.04_amd64.deb-bundle.tar

Install LibSSL otherwise it will fail.

wget http://archive.ubuntu.com/ubuntu/pool/main/o/openssl/libssl1.1_1.1.1f-1ubuntu2_amd64.deb
sudo dpkg -i libssl1.1_1.1.1f-1ubuntu2_amd64.deb

Install Libaio1 (Lol - AIO 😋) otherwise it will fail.

wget http://archive.ubuntu.com/ubuntu/pool/main/liba/libaio/libaio1_0.3.110-5_amd64.deb
sudo dpkg -i libaio1_0.3.110-5_amd64.deb

Install MySQL common.

dpkg -i mysql-common_8.0.27-1ubuntu20.04_amd64.deb

 Without the above being done, it errors out stating it needs MySQL common. If you try to include the package (./mysql-common....) with apt install, this fails due to it being the "last" package it installs and fails to understand when to install it.

Finally, install MySQL.

apt install --yes --no-install-recommends ./mysql-community-client_8.0.27-1ubuntu20.04_amd64.deb ./mysql-community-client-core_8.0.27-1ubuntu20.04_amd64.deb ./mysql-community-client-plugins_8.0.27-1ubuntu20.04_amd64.deb ./mysql-server_8.0.27-1ubuntu20.04_amd64.deb ./mysql-community-server_8.0.27-1ubuntu20.04_amd64.deb ./mysql-client_8.0.27-1ubuntu20.04_amd64.deb ./mysql-community-server-core_8.0.27-1ubuntu20.04_amd64.deb

Enter root password for password authentication or leave blank to just run mysql and login with terminal.

Choose "Use Strong Password...", as you won't be needing 5.x compatibility. 

Complete. Run mysql.

image.png

Networking With Hetzner - With and Without additional MAC's configured

IP's have been redacted in this configuration but you can evaluate based on your operations.

With doing this in debian, reverting over to netplan and disabling networking has been so much easier. ChatGPT and other help articles from stack overflow has been a major help in understanding networking as a whole. systemctl stop networking && systemctl disable networking. Install netplan.io with apt install netplan.io.

Configure it then reboot the system. Do not presume it will work unless you restart.

If you just need the netplan configuration:

network:
  version: 2
  renderer: networkd
  ethernets:
    enp41s0: # Network interface name from "ip a".
      dhcp4: no  # Turn off DHCP.
      addresses:
        - 148.x.x.242/29  # IP Address 1 with subnet.
        - 144.x.x.121/27  # IP Address 2 with subnet.
      gateway4: 148.x.x.241  # The gateway of the first subnet.
      nameservers:
        addresses:
          - 8.8.4.4 # Obvious enough.
          - 8.8.8.8 # Obvious enough.
      routes:
        - to: 144.x.x.96/27 # The "Identifier IP" for the subnet. The one before the gateway and two before the first usable.
          via: 144.x.x.97 # The gateway of this subnet.
          metric: 100 # The higher the metric, the lower it is in priority (Yes, stupid) - It means if 240 fails, it will make 144 prioritized.

If you need to specify MAC addresses:

network:
  version: 2
  renderer: networkd
  ethernets:
    enp41s0: # Network interface name from "ip a".
      dhcp4: no # Turn off DHCP.
      addresses:
        - 148.x.x.242/29 # IP Address 1 with subnet.
        - 144.x.x.121/27 # IP Address 2 with subnet.
      macaddress: xx:xx:xx:xx:xx:xx  # Specify the MAC address for 148.x.x.242 ONLY (It has to route somewhere for remaining traffic).
      routes:
        - to: 144.x.x.96/27 # The "Identifier IP" for the subnet. The one before the gateway and two before the first usable.
          via: 144.x.x.97 # The gateway of this subnet.
          metric: 100 # The higher the metric, the lower it is in priority (Yes, stupid) - It means if 240 fails, it will make 144 prioritized.
        - to: 148.x.x.240/29 # The "Identifier IP" for the subnet. The one before the gateway and two before the first usable.
          via: 148.x.x.241 # The gateway of this subnet.
          metric: 0 # The higher the metric, the lower it is in priority (Yes, stupid) - Not applicable here - 0 meaning normal priority.

-----------------------------------------------------------------------------------------------
    # If using multiple interfaces, add them here with their respective MAC addresses
    enp41s0:1:
      dhcp4: no
      addresses:
        - 144.x.x.121/27
      macaddress: yy:yy:yy:yy:yy:yy  # MAC address for 144.x.x.121

 

Running Lancache & Another Webserver On Top

If you are running the software Lancache and need to run another webserver on top such as nginx with ports 80 and 443, you need to do some configuration changes by serving your virtual host files on the public IP address while your Lancache speaks on your private internal IP's to your dedicated systems/LAN network.

In this example, I will use Lancache on 10.10.10.50 and the public IP as 123.123.123.123. What we need to do first is configure Lancache with the following inside the .env file:

image.png

Once this has been configured, head into the docker-compose.yml and update the 80 and 443 ports to be bound onto that particular bind of private IP.

image.png

Once completed, start up Lancache first as it needs port 80 and 443 in use for the Monolithic container using docker-compose up. Check that it was successful, then Ctrl+C, then docker-compose up -d.

We've now configured Lancache to work on that private LAN IP address. For all your virtual host configuration files now for nginx, you just need to specify another private LAN IP, or the public IP if serving accessible websites (if you aren't using HAProxy).

You can now restart nginx and it will not complain that the port is already in use as you've done this neat hack.

Keep in mind that if you have software that updates the nginx vhost files automatically such as TenantOS or VirtFusion, please keep in mind that you will probably need to repeat this again on their updates.

Installing Debian/Ubuntu & Then Disk Unmounts - Debootstrap Error

You're here because you've landed yourself at the error below, stating that it could not find the codename for the release during an installation of Ubuntu or Debian. This is because the CD unmounted during your ISO install, so after provisioning the disks, it's no longer there and cannot continue. Follow below to fix it

image.png

Press enter/escape and head back onto the installer page, then going to "Execute a shell".

Press continue.

You're now in the BusyBox shell. Type "blkid" and find the CD drive. It's got the name of the ISO.

In this example, we can see it has the type iso9660, so we know that's the ISO, and the mount is /dev/sr2. Now simply mount it! mount /dev/sr2 /cdrom.

Make sure it's /cdrom as this is what the installer is looking for.

Type exit then press back on install.

Job done!