sudo mkfs.ext4 /dev/<device name> sudo mkdir -p /var/lib/gravity echo -e "[Mount]\nWhat=/dev/<device name>\nWhere=/var/lib/gravity\nType=ext4\n[Install]\nWantedBy=local-fs.target" | sudo tee /etc/systemd/system/var-lib-gravity.mount sudo systemctl daemon-reload sudo systemctl enable var-lib-gravity.mount sudo systemctl start var-lib-gravity.mount
Hardware Prerequisites
To ensure the performance and stability of Anypoint Platform Private Cloud Edition (Anypoint Platform PCE), every node in your Anypoint Platform PCE environment must meet the requirements described in this topic.
Before you install Anypoint Platform PCE, your infrastructure team must review each of the following sections and verify that your environment meets the stated disk and device requirements. If needed, contact your MuleSoft representative for assistance. |
Memory and CPU Requirements
Component | Requirement |
---|---|
RAM |
32 GB |
CPU |
8 Cores |
If the stated requirements for Memory and CPU are not met, Anypoint Platform PCE installation will fail. |
For production environments, each node in your configuration must have the following dedicated devices. Except for the Docker device, all devices must be formatted either as xfs or ext4.
Component | Size | Minimum IOPS | Mount Point | Description |
---|---|---|---|---|
100 GB |
1500 |
/var/lib/gravity |
Stores system configuration and metadata, including databases and packages. Because package sizes can be large, it is important to estimate the minimum size requirements and allocate enough space as a dedicated device before installation. |
|
HDD 3 Application Data Device |
250 GB |
1500 |
/var/lib/data |
Stores application configuration and data. The amount of space required should be at least 250 GB, but can vary depending on your specific use case. It is important to estimate the minimum size requirements and allocate enough space as a dedicated device ahead of time. |
HDD 4 Docker Device |
100 GB |
3000 |
Not mounted |
Used by Docker’s Device Mapper storage driver. A minimum of 100Gb is required for the Device Mapper directory. Devices with 50Gb or less will experience degraded system performance or might not work at all. |
HDD 5 Etcd Device |
20 GB |
3000 |
/var/lib/gravity/planet/etcd |
Provides dedicated storage for the distributed database used for cluster coordination. |
/tmp (Installer) |
20 GB |
N/A |
||
/ (root) |
10 GB |
N/A |
You must have at least 10GB in your home directory to download and unzip the installer file. |
The space requirements listed in the previous table are based on average use. For environments with heavy traffic, validate the amount of space needed with your Customer Success Manager. |
System State Directory Device
The main purpose of the system state directory device is to store system configuration and metadata, for example, database and packages. As package sizes can be arbitrarily large, it is important to estimate the minimum size requirements and allocate enough space for the dedicated device.
Format this device either as xfs
or ext4
and mounted as /var/lib/gravity
.
The following shell snippet is provided as an example for mounting the system state directory device using systemd mount files and cannot be used as-is in production. You can also mount the system state directory device by using /etc/fstab
or some other method. Consult with your system administrator to determine the best way to meet this requirement. If you use the following example to guide this process, make sure to specify the correct device name in two places.
Application Data Device
The main purpose of the application data directory device is to store application configuration and data. The minimum amount of space required is 250GB, but your use case might require more. It is important to allocate enough space for the dedicated device.
Format this device either as xfs
or ext4
and mounted as /var/lib/data
.
The following shell snippet is provided as an example for mounting the application data device using systemd mount files, and cannot be used as-is in production. You can also mount the application data device using /etc/fstab
or some other method. Consult with your system administrator to determine the best way to meet this requirement. If you use the following example to guide this process, make sure to specify the correct device name in two places.
sudo mkfs.ext4 /dev/<device name> sudo mkdir -p /var/lib/data echo -e "[Mount]\nWhat=/dev/<device name>\nWhere=/var/lib/data\nType=ext4\n[Install]\nWantedBy=local-fs.target" | sudo tee /etc/systemd/system/var-lib-data.mount sudo systemctl daemon-reload sudo systemctl enable var-lib-data.mount sudo systemctl start var-lib-data.mount
Etcd Device
The main purpose of the etcd device is to provide dedicated storage for a distributed database used for cluster coordination. It does not require much space, 20GB is adequate.
Format this device either as xfs
or ext4
and mounted as /var/lib/gravity/planet/etcd
.
The following shell snippet is provided as an example for mounting the etcd device using systemd mount files, and cannot be used as-is in production. The end result of this step mounts the Etcd device to the path /var/lib/gravity/planet/etcd
. You can also mount the etcd device using /etc/fstab
or some other method. Consult with your system administrator to determine the best way to meet this requirement. If you use the following example to guide this process, make sure to specify the correct device name in two places.
sudo mkfs.ext4 /dev/<device name> sudo mkdir -p /var/lib/gravity/planet/etcd echo -e "[Mount]\nWhat=/dev/<device name>\nWhere=/var/lib/gravity/planet/etcd\nType=ext4\n[Install]\nWantedBy=local-fs.target" | sudo tee /etc/systemd/system/var-lib-gravity-planet-etcd.mount sudo systemctl daemon-reload sudo systemctl enable var-lib-gravity-planet-etcd.mount sudo systemctl start var-lib-gravity-planet-etcd.mount
Docker Device
Unless specified, Docker configuration defaults to the use of Device Mapper in loopback mode (using /dev/loopX
devices), which is not recommended for production. To configure Docker to use a dedicated device for Device Mapper storage driver, an unformatted device or a partition (/dev/sdd) can be provided during installation. This directory is automatically configured and set up for use.
Unformatted devices potentially usable for system directory / Device Mapper are automatically discovered by agents running on each node. Discovered devices are offered on a drop-down menu for configuration before the installation is started.
You can list unmounted devices with the following command:
lsblk --output=NAME,TYPE,SIZE,FSTYPE -P -I 8,9,202|grep 'FSTYPE=""'
Unmounted devices have an empty value in FSTYPE column. Devices with TYPE="part" are partitions on another device. The following command lists only specific device types:
Device type |
Description |
8 |
SCSI disk devices |
9 |
Metadisk (RAID) devices |
202 |
Xen virtual block devices (Amazon EC2) |
Reset Docker Device
Using Logical Volume Manager, you can group multiple physical volumes into a single storage volume (volume group) and then divide these into logical volumes. Physical volumes are either a whole device or a partition.
When a Docker device is in use by another logical volume or you want to manually reset a Docker device previously configured for Device Mapper, perform the following commands to reset the Docker device:
# In order to clean the Docker volumes, you might need to first reboot the server # to force clean down any handles to the Docker Device. # This allows you to perform the proper clean-up. # Identify whether the docker logical volume was created. sudo lvs | grep docker # If the volume exists, remove it. sudo lvremove docker # Identify whether the docker volume group was created. sudo vgs | grep docker # If the volume exists, remove it. sudo vgremove docker # Identify whether the docker physical volume was created. sudo pvs | grep docker # If the volume exists, remove it. sudo pvremove <PV Name> # Clean the docker device by replacing "{docker-device}" in the following command: sudo parted /dev/{docker-device} mklabel gpt -s
Verify Devices
To verify that all devices are set up the same way on all nodes, run the lsblk
command. Output from lsblk
should resemble the following example:
[root@ip-0-0-0-0 ~]# lsblk NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT xvda 202:0 0 90G 0 disk ├─xvda1 202:1 0 1M 0 part └─xvda2 202:2 0 90G 0 part / xvdb 202:16 0 100G 0 disk {Docker volume (not formatted)} xvdc 202:32 0 20G 0 disk /var/lib/gravity/planet/etcd xvdd 202:48 0 250G 0 disk /var/lib/data xvde 202:64 0 100G 0 disk /var/lib/gravity xvdf 202:64 0 20G 0 disk /tmp
Verify Disk IOPS
Depending on your hardware or virtualization provider, you can configure disk IOPS (I/O operations per second). Using the iops
command, verify available IOPS on each device:
$ sudo ./iops --time 2 /dev/xvdb /dev/xvdb, 107.37 GB, 32 threads: 512 B blocks: 1893.0 IO/s, 946.5 KiB/s ( 7.8 Mbit/s) 1 KiB blocks: 1354.8 IO/s, 1.3 MiB/s ( 11.1 Mbit/s) 2 KiB blocks: 1091.8 IO/s, 2.1 MiB/s ( 17.9 Mbit/s) 4 KiB blocks: 807.1 IO/s, 3.2 MiB/s ( 26.4 Mbit/s) 8 KiB blocks: 803.7 IO/s, 6.3 MiB/s ( 52.7 Mbit/s) 16 KiB blocks: 787.4 IO/s, 12.3 MiB/s (103.2 Mbit/s) 32 KiB blocks: 700.8 IO/s, 21.9 MiB/s (183.7 Mbit/s) 64 KiB blocks: 590.0 IO/s, 36.9 MiB/s (309.3 Mbit/s) 128 KiB blocks: 327.6 IO/s, 40.9 MiB/s (343.5 Mbit/s) ...
Load Balancer Requirements
Anypoint Platform PCE must run in a production environment with multiple servers. To distribute traffic among servers and to restrict access to specific ports, you must provide and configure a load balancer.
You can use any load balancer, including NGINX, but NGINX is not required.
If using the AWS Provisioner for Anypoint Private Cloud, you do not need to configure a separate load balancer. An Amazon ELB with the required configuration is created during installation. |
Load Balancer Configuration
You can configure your load balancer to use any method for distributing client requests, but in most contexts a round robin strategy is ideal. The load balancer should be accessible through an IP address that is available to all machines in your network.
Your load balancer must route the following TCP ports:
Load Balancer Port | Instance Port | Internal Usage |
---|---|---|
|
|
HTTP redirects to HTTPs. |
|
|
HTTPS port |
|
|
WebSocket port for Runtimes to connect |
|
|
Ops Center access port |
In each case, your load balancer must listen on the load balancer port and redirect incoming requests to the instance port. Your Anypoint Platform PCE installation includes an internal NGINX server that listens on each of the configured instance ports, and then performs the action listed in the Internal Usage column.
Your load balancer should poll the address HTTP:10248/healthz
to run a health check on your platform servers and confirm that they are accessible.
Do not configure SSL certificates in your load balancer. TLS termination is handled by the platform with the certificates configured using Access Management. See Configure Anypoint Platform PCE. |
Implement a Load Balancer Using NGINX
-
Enable stream block in your
/etc/nginx/nginx.conf
file by referencing all the configuration properties defined in/etc/nginx/stream.d/*
.stream { include /etc/nginx/stream.d/*.conf; }
-
Delete the
default.conf
file from/etc/nginx/conf.d
-
Create a folder named
/etc/nginx/stream.d
and in it create a file namedonprem.conf
-
Paste the following content in
onprem.conf
server { listen 80; proxy_pass 0.0.0.0:30080; } server { listen 443; proxy_pass 0.0.0.0:30443; } server { listen 8889; proxy_pass 0.0.0.0:30889; } server { listen 9500; proxy_pass 0.0.0.0:32009; } server { listen 9501; proxy_pass 0.0.0.0:30083; }
-
Enable the default ports on SELinux before starting the NGINX. Run the following commands one by one:
semanage port -a -t http_port_t -p tcp 8889 semanage port -a -t http_port_t -p tcp 9500 semanage port -a -t http_port_t -p tcp 9501
-
Start NGINX:
service nginx restart