Adding a Worker Node - GCP

Overview

This procedure is for adding a worker node to an existing TOS Aurora cluster on the GCP platform. If you have not yet installed TOS Aurora, on the primary data node, start with Clean Install of TOS Aurora on GCP. For all other installation paths such as upgrade or other platforms, see the menu for the appropriate procedure.

You do not need to install TOS on the worker nodes.

Prerequisites

  • This procedure must be performed by an experienced Linux administrator with knowledge of network configuration.

  • To ensure optimal performance and reliability, the required resources need to be allocated exclusively to TOS. If resources become unavailable, this will affect TOS performance. Do not oversubscribe resources.

  • You cannot use IP Tables. All IP tables rules will be flushed when adding the node.

  • Your primary data node must also be deployed on GCP.

  • You must know the resources you will need - CPU cores, RAM, disk space and the load-model parameter, provided by your account team based on the procedure Calculate resources - clean install.

  • (On-premises deployments only) The node's network IP must be on the same subnet as the cluster primary VIP.

  • Give the node a unique hostname in the cluster - use the command below, replacing <mynode> with your preferred name:

    • [<ADMIN> ~]$ sudo hostnamectl set-hostname <mynode>
    sudo hostnamectl set-hostname <mynode>

Operating System Requirements

  • OS distribution:

    • Red Hat Enterprise Linux 8.10

    • Rocky Linux 8.10

  • Disks: Select a storage type of SSD. Take into consideration that TOS requires 7,500 IOPS and the throughput expected will average 250MB/s with bursts of up to 700MB/s.

  • You need to configure a separate partition for /opt,, and the OS disk needs at least 300 GB of available storage. The /opt partition will contain your data, which will increase over time. Most of your available disk space should be allocated to this partition and the minimum is determined by the load model parameter (small, medium, large) provided by your account team.

    Minimum sizes for all partitions:

    • OS disk: 150 GB

    • /opt/ (small): 70 GB

    • /opt/ (medium): 70 GB

    • /opt/ (Large): 70 GB

    • etcd: N/A

    *Small, medium and large refer to the load model parameter provided by your account team.

    We recommend allocating /opt partition all remaining disk space after you have partitioned the OS disk and moved etcd to a separate disk.

  • The kernel must be up-to-date

  • SELinux must be disabled

  • Language: en-US

  • You must have permissions to execute TOS CLI commands located in directory /usr/local/bin/tos and to use sudo if necessary.

  • To run TOS CLI commands without specifying the full path (/usr/local/bin/tos), your environment path must be modified accordingly.

  • The server timezone must be set.

Network Requirements

  • Tufin Orchestration Suite must only be installed in an appropriately secured network and physical location. Only authorized users should be granted access to TOS products and the operating system on the server.

  • You must allow access to required Ports and Services.

  • All TOS nodes need to be on the same subnet and layer 2 network that supports ARP (address resolution protocol).

  • All TOS nodes should have network latency of under 1ms.

  • Network configurations for your interface must be set to manual IPv4 with gateway and DNS Servers set to the IPs used by your organization.

    The system will use a reverse DNS lookup (PTR record) to resolve the DNS IP addresses with the domain name during the TOS installation. Therefore you have to add these PTR records to the DNS server. If you do not, the TOS installation will fail.

  • Allocate a 24-bit CIDR subnet for the Kubernetes service network and a16-bit CIDR subnet for the Kubernetes pods network (10.244.0.0/16 is used by default).

    The pods and services networks must be inside the following private networks: 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16. In addition, ensure that the dedicated CIDR for the service network and pods network don't overlap with each other, and:

    • The physical addresses of your TOS Aurora servers (see below)

    • Your primary VIP, Syslog VIP or external load balancer IP (see below)

    • Any other subnets communicating with TOS or with TOS nodes

  • If a proxy is configured on your system make sure this network is excluded.

  • You must have available the following dedicated IP addresses:

    • For on-premise deployments, a primary VIP that will serve as the external  IP address used to access TOS Aurora from your browser. The primary VIP will not be needed in the installation of the operating system, except in the final step - the installation command.
    • The physical network IP address of the first network interface used by the administrator for CLI commands. This is the IP address you will use in most steps of the procedure.

    • If additional nodes are subsequently added to the cluster, each node will require an additional dedicated physical network IP address.

    • Additional syslog VIPs can be allocated as needed.

    • The VIP, all node physical network IP addresses and all syslog VIPs must be on the first network interface.

    • Make sure your first physical interface is correctly configured and all other interfaces are not on the same network.

      To find the first network interface, run the following command:

      [<ADMIN> ~]$ sudo /opt/tufinos/scripts/network_interface_by_pci_order.sh | awk -F'=' '/NET_IFS\[0\]/ { print $NF }'
      sudo /opt/tufinos/scripts/network_interface_by_pci_order.sh | awk -F'=' '/NET_IFS\[0\]/ { print $NF }'

      Otherwise network errors such as connectivity failures and incorrect traffic routing might occur.

Procedure

Before you proceed, read and understand Prerequisites - this may prevent unexpected failures.

  1. For additional help, refer to the official documentation Create and start a VM instance.

    1. In the Google Cloud console, navigate to Compute Engine > VM instances

    2. Click +CREATE INSTANCE.

    3. Configure:

      • Name:

      • Region/Zone: As appropriate

      • Machine family: GENERAL-PURPOSE

      • Series: E2

      • Machine type: Custom

      • Cores / Memory: According to the load-model parameter value received from your account team.

    4. Under Boot Disk, click CHANGE.

    5. Configure the boot disk.

      • Make sure tab Public Images is selected.

      • Operating system: Red Hat Enterprise Linux or Rocky Linux

      • Version:

        • Red Hat Enterprise Linux 8.10

        • Rocky Linux 8.10

      • Boot disk type, SSD persistent disk

      • Size, The value received from your account team.

    6. Click Select. The boot disk settings will be saved.

    7. Select Advanced options.

    8. Select Networking.

    9. Configure Networking:

      • Network tags: default-allow-ssh allow-nginx-ingress

      • Hostname: Optional

    10. (Optional) Host name: Type a host name or use the default device value.

    11. Click CREATE.

  2. Add the VM to the Instance Group

    For more information about managing instance groups, see the vendor documentation Group unmanaged VMs together.

    Leave default values for all parameters not mentioned.

    1. Navigate to Instance groups.

    2. Select the instance group that contains the existing VMs in the cluster.

    3. Configure:

      • VM Instances: Add the new VM created previously

    4. Click SAVE.

  3. For more information, see the vendor documentation Set up the load balancer.

    Leave default values for all parameters not mentioned.

    1. Navigate to Load balancing - Network services.

    2. Select the load balancer created previously for the primary data node

    3. Click Backend configuration:

    4. Enter values:

      • Backends: SELECT EXISTING INSTANCES

      • VM instances: Add the new VM created previously

    5. Click DONE.

  4. To connect via ssh you can add an existing ssh key to GCP via UI or use the gcloud utilily to generate key and add it.

    To generate a key using gcloud:

    1. Download gcloud to your workstation.

    2. Run gcloud auth login command to login to your account. Variable <PROJECT_ID> can be set once globally to avoid entering it each time.

      gcloud config set project "<PROJECT_ID>"
      gcloud config set project "<PROJECT_ID>"
      gcloud compute ssh --zone "<ZONE>" "<VM_INSTANCE_NAME>" 
      gcloud compute ssh --zone "<ZONE>" "<VM_INSTANCE_NAME>"

  5. If not done already, set up partitions according to the Prerequisites.

    1. If you are not currently logged in as user root, do so now.

      [<ADMIN> ~]$ su -
      su -

    2. If you want to change the host name or IP of the machine, do so now. Once TOS Aurora has been installed, changing the host name or IP address will require reinstalling - see Changing IP Address/Host Names. To change the host name, use the command below, replacing <mynode> with your preferred name.

      [<ADMIN> ~]# hostnamectl set-hostname <mynode>
      hostnamectl set-hostname <mynode>

    3. Modify the environment path to run TOS CLI commands without specifying the full path (/usr/local/bin/tos). 

      [<ADMIN> ~]# echo 'export PATH="${PATH}:/usr/local/bin"' | sudo tee -a /root/.bashrc > /dev/null
      echo 'export PATH="${PATH}:/usr/local/bin"' | sudo tee -a /root/.bashrc > /dev/null

    4. Synchronize your machine time with a trusted NTP server. Follow the steps in Configuring NTP Using Chrony.

    5. Configure the server timezone.

      [<ADMIN> ~]# timedatectl set-timezone <timezone>
      timedatectl set-timezone <timezone>

      where <timezone> is in the format Area/Location. Examples: America/Jamaica, Hongkong, GMT, Europe/Prague. List the time-zone formats that can be used in the command.

      [<ADMIN> ~]# timedatectl list-timezones
      timedatectl list-timezones

    6. Upgrade the kernel:

      [<ADMIN> ~]# dnf upgrade
      dnf upgrade

    7. Disable SELinux:

      • If file /etc/selinux/config exists, edit and change the value of SELINUX to disabled:

        SELINUX=disabled
      • If the file doesn't exist or SELINUX is already set to disabled, do nothing.
    8. Reboot the machine and log in.

    9. Install Wireguard. This is needed to encrypt communication between nodes (machines) within the cluster. The wireguard version must match the operating version you are installing.

      10. [<ADMIN> ~]# sudo yum install https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm
      sudo yum install https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm
      [<ADMIN> ~]# sudo yum install kmod-wireguard wireguard-tools
      sudo yum install kmod-wireguard wireguard-tools
    10. Reboot the machine and log in.

    11. Install tmux and rsync:

      [<ADMIN> ~]# dnf install -y rsync tmux
      dnf install -y rsync tmux

    12. Disable the firewall:

      [<ADMIN> ~]# systemctl stop firewalld
      systemctl stop firewalld
      [<ADMIN> ~]# systemctl disable firewalld
      systemctl disable firewalld

    13. Create the TOS Aurora load module configuration file /etc/modules-load.d/tufin.conf. Example using vi:

      [<ADMIN> ~]# vi /etc/modules-load.d/tufin.conf
      vi /etc/modules-load.d/tufin.conf

    14. Specify the modules to be loaded by adding the following lines to the configuration file created in the previous step. The modules will then be loaded automatically on boot.

      br_netfilter
      wireguard
      overlay
      ebtables
      ebtable_filter
      br_netfilter wireguard overlay ebtables ebtable_filter

    15. Load the above modules now:

      [<ADMIN> ~]# cat /etc/modules-load.d/tufin.conf |xargs modprobe -a 
      cat /etc/modules-load.d/tufin.conf |xargs modprobe -a

      Look carefully at the output to confirm all modules loaded correctly; an error message will be issued for any modules that failed to load.

    16. Check that Wireguard has loaded correctly. 

      [<ADMIN> ~]# lsmod |grep wireguard
      lsmod |grep wireguard

      The output will appear something like this:

      wireguard              201106  0
ip6_udp_tunnel         12755  1 wireguard
udp_tunnel             14423  1 wireguard

      If Wireguard is not listed in the output, contact support.

    17. Create the TOS Aurora kernel configuration file /etc/sysctl.d/tufin.conf. Example using vi:

      [<ADMIN> ~]# vi /etc/sysctl.d/tufin.conf
      vi /etc/sysctl.d/tufin.conf

    18. Specify the kernel settings to be made by adding the following lines to the configuration file created in the previous step. The settings will then be applied on boot. 

      net.bridge.bridge-nf-call-iptables = 1
      fs.inotify.max_user_watches = 1048576
      fs.inotify.max_user_instances = 10000
      net.ipv4.ip_forward = 1
      net.bridge.bridge-nf-call-iptables = 1 fs.inotify.max_user_watches = 1048576 fs.inotify.max_user_instances = 10000 net.ipv4.ip_forward = 1

    19. Apply the above kernel settings now:

      [<ADMIN> ~]# sysctl --system
      sysctl --system
    For maximum security, we recommend only installing official security updates and security patches for your Linux distribution, as well as the RPMs specifically mentioned in this section.

    1. Log in to the primary data node.

    2. On the primary data node:

      [<ADMIN> ~]$ sudo tos cluster node add --role=worker
      sudo tos cluster node add --role=worker

      On completion, a new command string is displayed, which you will need to run on the new node within 30 minutes. If the allocated time expires, you will need to repeat the current step.

    3. Copy the command string to the clipboard.

    4. Log in to the new node.

    5. On the new node, paste the command string copied previously and run it. If the allocated time has expired, you will need to start from the beginning.
    6. Verify that the node was added by running sudo tos cluster node list on the primary data node.

    1. On the primary data node, check the TOS status.

      [<ADMIN> ~]$ tos status
      tos status

    2. In the output, check if the System Status is Ok and all the items listed under Components appear as Ok. If this is not the case, contact Tufin Support.

      3. Example output for a central cluster data node:

      [<ADMIN> ~]$ tos status         
[Mar 28 13:42:09]  INFO Checking cluster health status           
TOS Aurora
Tos Version: 24.2 (PRC1.1.0)

System Status: "Ok"
            
Cluster Status:
   Status: "Ok"
   Mode: "Multi Node"

Nodes
  Nodes:
  - ["node1"]
    Type: "Primary"
    Status: "Ok"
    Disk usage:
    - ["/opt"]
      Status: "Ok"
      Usage: 19%
  - ["node3"]
    Type: "Worker Node"
    Status: "Ok"
    Disk usage:
    - ["/opt"]
      Status: "Ok"
      Usage: 4%

registry
  Expiration ETA: 819 days
  Status: "Ok"

Infra
Databases:
- ["cassandra"]
  Status: "Ok"
- ["kafka"]
  Status: "Ok"
- ["mongodb"]
  Status: "Ok"
- ["mongodb_sc"]
  Status: "Ok"
- ["ongDb"]
  Status: "Ok"
- ["postgres"]
  Status: "Ok"
- ["postgres_sc"]
  Status: "Ok"

Application
Application Services Status OK
Running services 50/50

Remote Clusters
Number Of Remote Clusters: 2
  - ["RC"]
     Connectivity Status:: "OK:"
  - ["RC2"]
     Connectivity Status:: "OK"

  Backup Storage:
  Location: "Local
s3:http://minio.default.svc:9000/velerok8s/restic/default "
  Status: "Ok"
  Latest Backup: 2024-03-23 05:00:34 +0000 UTC

      Example output for a remote cluster data node:

      [<ADMIN> ~]$ tos status         
[Mar 28 13:42:09]  INFO Checking cluster health status           
TOS Aurora
Tos Version: 24.2 (PRC1.0.0)

System Status: "Ok"
            
Cluster Status:
   Status: "Ok"
   Mode: "Single Node"

Nodes
  Nodes:
  - ["node2"]
    Type: "Primary"
    Status: "Ok"
    Disk usage:
    - ["/opt"]
      Status: "Ok"
      Usage: 19%
  
registry
  Expiration ETA: 819 days
  Status: "Ok"

Infra
Databases:
- ["mongodb"]
  Status: "Ok"
- ["postgres"]
  Status: "Ok"

Application
Application Services Status OK
Running services 16/16

  Backup Storage:
  Location: "Local
s3:http://minio.default.svc:9000/velerok8s/restic/default "
  Status: "Ok"
  Latest Backup: 2024-03-23 05:00:34 +0000 UTC

After the node is added, we recommend stopping tos and then starting it to enhance the node's performance. This will require downtime.