In the blog article () about the ArcGIS 10.1 for Server architecture, we know that the GIS Server is equal to each other as part of the website architecture of ArcGIS 10.1 for Server, clusters can be performed between multiple GIS servers. In this blog, we will introduce how to build a Server cluster on RHEL.
Note: The operating system environment is very important when a cluster is created on Linux. Why? First, it is not easy to install software on Linux, and package shortage often occurs. If you are not familiar with Linux, permissions may also become a problem. Linux has strict control permissions on files, this is much more rigorous than Windows. Finally, some servers are set up. To build a cluster on Linux, each participating machine must access each other and have read and write permissions on the data directory, this requires the DNS server and NFS (not necessarily NFS, but other servers can be used) servers.
========================================================== ==================
Install ArcGIS 10 on Linux:
Install ArcGIS 10_01 on Linux to install Linux Virtual Machine and yum source configuration
Install ArcGIS 10_02 on Linux and Oracle11gR2
Install ArcGIS 10_03 on Linux to install ArcSDE 10
Install ArcGIS 10_04 on Linux to install ArcGIS Server 10
Install ArcGIS 10.1 _ 01 on Linux to install ArcGIS Server 10.1
========================================================== ==================
Basic settings
1. Firewall disabled
Disable firewall for machine A and machine B
2. arcgis users and groups
The arcgis users and groups we set here are: arcgis (note that the same account and password are used to install software on the two machines)
Check environment requirements
Installing software on Linux is indispensable for system check. Esri provides us with a script CheckPacagePrerequisites. sh to check whether the system meets the conditions.
Run sh CheckPacagePrerequisites. sh and select y to indicate that the conditions are met:
If the following error message is not displayed:
Arcgis soft nofile 65535
Arcgis hard nofile 65535
Arcgis soft nproc 25059
Arcgis hard nproc 25059
File and process launch
ArcGIS Server is a data-intensive Server product whose data format is composed of hundreds of thousands of files. When a large number of systems are used, thousands of files may run at any time, if there are not enough file handles and processes, the request may fail to start, resulting in system downtime. The actual number of file handles and processes must be set based on different data and instance numbers. Here we set 65535 file handles and 25059 processes, to ensure that the system continues to run.
Log on as a root user and run the command gedit/etc/security/limits. conf on the terminal. Edit the following content:
Arcgis soft nofile 65535
Arcgis hard nofile 65535
Arcgis soft nproc 25059
Arcgis hard nproc 25059
Save and exit, log out, and log on again to make the change take effect. Run ulimit-Hn and ulimit-Hu to view the output.
Create two folders under the/usr directory
Arcgis and test. arcgis is the installation directory, test is the configuration directory, and the test directory must be a shared directory.
Set the 777 permission for the arcgis directory and test directory, or change it to the arcgis user. Our practice here is to change both directories to belong to arcgis users.
Machine name Modification
First, check the machine name in the terminal and enter hostname: the output is localhost. localdomain
Run gedit/etc/sysconfig/network to edit
Input: HOSTNAME = set the machine name here.
Note: The specified Machine name is used in DNS configuration.
DNS Server Configuration
Due to the internal mechanism of Server10.1, you only need to configure the machine name and IP address to the DNS before configuring the cluster, and also need to perform forward and reverse DNS resolution. Forward resolution refers to domain name resolution by domain name resolution IP address, reverse resolution by IP Address Resolution domain name.
Specific practice: directly edit two configuration files under the/var/named directory of the DNS server: hosts and another configuration file for reverse resolution, and add the IP address and machine name to the configuration file respectively. Restart the DNS service to make the change take effect.
Shared directory Configuration
Because two machines (server88 and server89) are used as clusters, You need to configure the shared directory. The following is the configuration method:
Server89 machine Configuration:
1) first disable the Firewall
2) Enter:
/Usr/arcgisserver 192.168.200.88 (rw, sync, no_root_squash)
Rw indicates read and write permissions, synv indicates Disk Synchronization, and no_root_squash indicates root permission if the directory is mounted as root.
3) restart the nfs service:
Service nfs restart
Run service nfs status to view the running status
4) restart the nfslock service:
Service nfslock restart
5) restart the rpcbind (formerly called portmap) service.
Service rpcbind restart
Server 88 machine Configuration:
1) first disable the Firewall
2) restart the rpcbind service (formerly called portmap ).
Service rpcbind restart
3) Run mount-t nfs 192.168.200.89:/arcgisserver/usr/arcgisserver to view the output. So far, the nfs file sharing directory has been created. You can create a file under the/usr/arcgisserver directory of 88, and check whether the file is successful under the/usr/arcgisserver directory of 89.
JDK Configuration
To configure the cluster, you must install Webadaptor for request forwarding. the java version is used. Therefore, you need to configure the JDL environment variable. The specific configuration is not detailed.
Installation and cluster configuration
Mount the installation disk file, log on to arcgis and enter the installation directory, and run the command to install the disk. The following describes the general installation process.
Start installation:
Select the/usr/arcgis directory we configured earlier:
Installing: