Difference between revisions of "Cloud"

From Hacksburg Wiki
Jump to navigation Jump to search
(Windows 10)
(Notes)
 
(3 intermediate revisions by the same user not shown)
Line 103: Line 103:
 
   Add-VpnConnection -Name "Hacksburg" -ServerAddress "cloud.hacksburg.org" –TunnelType IKEv2 -EncryptionLevel Required -AuthenticationMethod EAP -SplitTunneling –AllUserConnection
 
   Add-VpnConnection -Name "Hacksburg" -ServerAddress "cloud.hacksburg.org" –TunnelType IKEv2 -EncryptionLevel Required -AuthenticationMethod EAP -SplitTunneling –AllUserConnection
 
   Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.15.0/24 -PassThru -AllUserConnection
 
   Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.15.0/24 -PassThru -AllUserConnection
 +
  Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.17.0/24 -PassThru -AllUserConnection
 
   Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.18.0/24 -PassThru -AllUserConnection
 
   Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.18.0/24 -PassThru -AllUserConnection
  
Line 178: Line 179:
 
  # systemctl start strongswan
 
  # systemctl start strongswan
 
  # swanctl -i -c Hacksburg
 
  # swanctl -i -c Hacksburg
 +
 +
== Software Tools ==
 +
 +
You can remote in to Hacksblade to use SprutCAM and Lightburn:
 +
 +
1. Install NoMachine (https://www.nomachine.com/). When proceeding through the instructions, do not install the server. If you already have NoMachine, no need to install it again.
 +
 +
1. Connect to VPN.
 +
 +
2. Open NoMachine.
 +
 +
3. In the search box, connect to "192.168.15.194".
 +
 +
4. Use your username and password. You will be prevented from logging in if someone else is using the computer.
 +
 +
5. '''Important''': Make sure to unselect "Change the server resolution to match the client when I connect". If you choose this, you can mess up the remote desktop for other users.
 +
 +
6. You are now presented with the Windows login. Login with the Hacksburg account, '''not your own.'''
 +
 +
7. Do your work.
 +
 +
8. Log out of the Hacksburg account using the start menu.
 +
 +
9. Close the NoMachine connection by clicking on the top right corner, choosing the power button, and clicking "Exit the Player".
  
 
== Notes ==
 
== Notes ==
  
 
https://pve.proxmox.com/wiki/ZFS:_Switch_Legacy-Boot_to_Proxmox_Boot_Tool#Switching_to_proxmox-boot-tool
 
https://pve.proxmox.com/wiki/ZFS:_Switch_Legacy-Boot_to_Proxmox_Boot_Tool#Switching_to_proxmox-boot-tool
 +
 +
https://sleeplessbeastie.eu/2017/03/06/how-to-use-hp-command-line-array-configuration-utility/
 +
 +
http://www.datadisk.co.uk/html_docs/redhat/hpacucli.htm
 +
 +
=== HP Server, Drive Replacement ===
 +
 +
# Remote into the server: <code>ssh root@192.168.15.2</code>
 +
# Confirm with ZFS that a drive has failed: <code>zpool status rpool</code>. One of the drives should be marked failed/offline/removed, something like that.
 +
# Confirm on the front of the server that a drive has failed: look for the red light. Note the number of the drive that has failed.
 +
# Open up the RAID controller utility: <code>hpacucli</code>.
 +
# You are now in the command line for the RAID controller utility. Type <code>ctrl slot=0 show config</code>. This will take a while, and should confirm that one of your logicaldrives has failed. The logicaldrive number should match the numbered bay on the front of the server.
 +
# Physically swap the drive: push the red button on the '''correct''' drive, pull it out, and push in the new drive until it clicks. '''WARNING: pulling the wrong drive at this moment will result in zfs redundancy failure and data loss.''' If you're really scared, shut down the computer first, and don't start it back up until you know you've swapped the correct drive.
 +
# Now run <code>ctrl slot=0 ld <NUMBER> modify reenable forced</code>, where <code><NUMBER></code> is the bay number of the (previously) failed drive. This tells the RAID controller that everything's fine and it should just carry on with the new disk.
 +
# Type <code>exit</code> to get out of the array configuration utility.
 +
# Run <code>zpool status rpool</code> again, and now one of the drives should definitely be marked as <code>REMOVED</code>. Note the (very long) name of that drive in the left column.
 +
# Replace the disk in ZFS: <code>zpool replace rpool <NAME> /dev/disk/by-id/<NAME-2></code>, where <code><NAME-2></code> is the same name as the drive, without any <code>-part3</code> suffix. For example, if <code><NAME></code> is <code>scsi-300320938420934-part3</code>, <code><NAME-2></code> should be <code>scsi-300320938420934</code>. I think this would be <code>-part9</code> if a previously replaced disk failed. The exact number is not important.
 +
# ZFS is now rebuilding (resilvering) the failed disk. The server may be slow for the next several hours. You can run <code>zpool status</code> to check on the progress. It's remaining time estimation is wildly too low, don't rely on it.

Latest revision as of 14:38, 8 September 2024

Ever wanted to learn about cloud computing, but were worried about monthly costs or confusing product options?

Wanted something new to learn that's stay-at-home compatible?

Hacksburg has computing resources on-site which members can use from anywhere.

What Hacksburg Offers

  • Member Virtual Machines
    • 2 cores, 6GiB RAM, 64GiB hard drive space
    • Private IP address on Hacksburg's network. Laptops, Raspberry Pis, etc. can all communicate directly with each other.
    • Public IPv6, so anyone on the Internet can connect to your server (subject to our firewall rules).
    • Gigabit Internet connection
    • *.<name>.cloud.hacksburg.org DNS wildcard
    • HTTP(S) for DNS wildcard to your server(s)
  • Colocation (bring your own server/Pi/etc.)
    • Subject to power limits (please don't blow the breaker)
    • Colocated servers also on Hacksburg network
  • VPN
    • Access your projects on Hacksburg's network from anywhere.

Hacksburg wants all members to explore computational projects using our resources. If there's something you think we can offer that isn't shown here, come to an open meeting or email board@hacksburg.org to discuss.

Want access?

  • To get a virtual machine, make sure you're an active Hacksburg member, then email board@hacksburg.org
    • VPN access will only be given if you are a board member, have a VM, or have some other reason. Email board@hackburg.org to request VPN access.

Hacksburg Physical Hardware

  • Main Server
    • HP ProLiant DL370 G6
    • 2x Xeon E5530, 8 Core / 16 Thread
    • 48GB of memory
    • 4x Gigabit Ports
    • 2x 10G SFP+ cages
    • Fibre channel
    • 8x 146GB 10K SAS in RAIDZ1 configuration (Total space reported: 1.06TB)
    • Runs ProxMox VM hosting software
  • Second Server
    • IBM System x3650 M3
    • 2x Xeon X5675, 12 Core / 24 Thread
    • 96GB of memory
    • 10x Gigabit Ports
    • 2x Fibre channel
    • Hard Drives TBD
    • Will run ProxMox VM hosting software
  • Sun Servers 1 and 2
    • SunFire X4200
    • Do not turn on the Sun Servers because they are unacceptably loud.
  • Switch
    • tp-link 8-port Gigabit switch, TL-SG108

Hacksburg Network Layout

  • pfSense Router
    • Internet comes in here
    • 192.168.15.1
  • Main Server
    • ProxMox web interface for the main server. This is where you go to interact with your VMs.
    • How to access: https://cloud.hacksburg.org
    • Where it's located: 192.168.15.2
  • Docker Host
    • Runs NGINX reverse proxy. All HTTP(S) traffic at Hacksburg runs through this NGINX reverse proxy. Members can request forwarding to their containers.
    • 192.168.15.3
    • Containers:
  • OctoPrint

VPN

VPN allows direct access to the Hacksburg network from anywhere else that has access to the public Internet. Email board@hacksburg.org to get your username and password.

Connection Instructions

Windows 10

These instructions may also work for Windows 8.1, 7, Vista, etc.

1. Download the Hacksburg certificate: http://hacksburg.org/Hacksburg+CA.crt.

2. Right-click and choose "Install certificate".

3. Choose "Local Machine" (second option) and click Next.

4. Choose "Place all certificates in the following store" (second option).

5. In the window that comes up, choose "Trusted Root Certification Authorities" (second option). Click OK.

6. Click Next.

7. Click Finish.

8. It should say that importing the certificate was successful. Click OK.

9. Open an administrator PowerShell and run the following commands:

 Add-VpnConnection -Name "Hacksburg" -ServerAddress "cloud.hacksburg.org" –TunnelType IKEv2 -EncryptionLevel Required -AuthenticationMethod EAP -SplitTunneling –AllUserConnection
 Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.15.0/24 -PassThru -AllUserConnection
 Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.17.0/24 -PassThru -AllUserConnection
 Add-VpnConnectionRoute -ConnectionName "Hacksburg" -DestinationPrefix 192.168.18.0/24 -PassThru -AllUserConnection

You should now have a VPN named Hacksburg. Click the network icon in taskbar tray, click Hacksburg, and connect with the provided username and password.

Linux (and friends)

Setup

Install strongswan. If you use NetworkManager (Ubuntu), you can use these instructions [1] and fill in the following values:

VPN Server IP: 216.252.204.76

VPN Username: Email board@hacksburg.org to request

VPN Password: Email board@hacksburg.org to request


If you do not use NetworkManager, add the following rows to your ipsec.conf file. On Arch Linux, this file is located at /etc/ipsec.conf:

conn Hacksburg
        keyexchange=ikev2
        dpdaction=clear
        dpddelay=30s
        eap_identity="<USERNAME>"
        leftauth=eap-mschapv2
        left=%defaultroute
        leftsourceip=%config
        right=cloud.hacksburg.org
        rightauth=pubkey
        rightsubnet=0.0.0.0/0
        rightid=%any
        rightca=<CACERT>
        type=tunnel
        auto=add

Where <CACERT> is replaced by the path to this file: http://hacksburg.org/Hacksburg+CA.crt and <USERNAME> is replaced by the username issued to you by emailing board@hacksburg.org. Then, create the file /etc/ipsec.secrets and include the following line:

<USERNAME> : EAP "<PASSWORD>"

Where the username and password are issued to you by emailing board@hacksburg.org. Then, run

# ipsec auto --add Hacksburg

Then, make sure /etc/xl2tpd/xl2tpd.conf has the following contents:

lac Hacksburg]
lns = 216.252.204.76
ppp debug = yes
pppoptfile = /etc/ppp/options.l2tpd.client
length bit = yes

And /etc/ppp/options.l2tpd.client has the following contents:

ipcp-accept-local
ipcp-accept-remote
refuse-eap
require-mschap-v2
noccp
noauth
idle 1800
mtu 1410
mru 1410
defaultroute
usepeerdns
debug
connect-delay 5000
name <USERNAME>
password <PASSWORD>

Again using the username and password you received from the board.

Start VPN

You can start the VPN by running:

# systemctl start strongswan
# swanctl -i -c Hacksburg

Software Tools

You can remote in to Hacksblade to use SprutCAM and Lightburn:

1. Install NoMachine (https://www.nomachine.com/). When proceeding through the instructions, do not install the server. If you already have NoMachine, no need to install it again.

1. Connect to VPN.

2. Open NoMachine.

3. In the search box, connect to "192.168.15.194".

4. Use your username and password. You will be prevented from logging in if someone else is using the computer.

5. Important: Make sure to unselect "Change the server resolution to match the client when I connect". If you choose this, you can mess up the remote desktop for other users.

6. You are now presented with the Windows login. Login with the Hacksburg account, not your own.

7. Do your work.

8. Log out of the Hacksburg account using the start menu.

9. Close the NoMachine connection by clicking on the top right corner, choosing the power button, and clicking "Exit the Player".

Notes

https://pve.proxmox.com/wiki/ZFS:_Switch_Legacy-Boot_to_Proxmox_Boot_Tool#Switching_to_proxmox-boot-tool

https://sleeplessbeastie.eu/2017/03/06/how-to-use-hp-command-line-array-configuration-utility/

http://www.datadisk.co.uk/html_docs/redhat/hpacucli.htm

HP Server, Drive Replacement

  1. Remote into the server: ssh root@192.168.15.2
  2. Confirm with ZFS that a drive has failed: zpool status rpool. One of the drives should be marked failed/offline/removed, something like that.
  3. Confirm on the front of the server that a drive has failed: look for the red light. Note the number of the drive that has failed.
  4. Open up the RAID controller utility: hpacucli.
  5. You are now in the command line for the RAID controller utility. Type ctrl slot=0 show config. This will take a while, and should confirm that one of your logicaldrives has failed. The logicaldrive number should match the numbered bay on the front of the server.
  6. Physically swap the drive: push the red button on the correct drive, pull it out, and push in the new drive until it clicks. WARNING: pulling the wrong drive at this moment will result in zfs redundancy failure and data loss. If you're really scared, shut down the computer first, and don't start it back up until you know you've swapped the correct drive.
  7. Now run ctrl slot=0 ld <NUMBER> modify reenable forced, where <NUMBER> is the bay number of the (previously) failed drive. This tells the RAID controller that everything's fine and it should just carry on with the new disk.
  8. Type exit to get out of the array configuration utility.
  9. Run zpool status rpool again, and now one of the drives should definitely be marked as REMOVED. Note the (very long) name of that drive in the left column.
  10. Replace the disk in ZFS: zpool replace rpool <NAME> /dev/disk/by-id/<NAME-2>, where <NAME-2> is the same name as the drive, without any -part3 suffix. For example, if <NAME> is scsi-300320938420934-part3, <NAME-2> should be scsi-300320938420934. I think this would be -part9 if a previously replaced disk failed. The exact number is not important.
  11. ZFS is now rebuilding (resilvering) the failed disk. The server may be slow for the next several hours. You can run zpool status to check on the progress. It's remaining time estimation is wildly too low, don't rely on it.