Directory Server 6 HA with Sun Cluster

本文档详细介绍了如何使用SunCluster 3.1或更高版本在Solaris 9或10系统上为SPARC、x86及x64平台安装DirectoryServer。通过Java ES安装程序进行安装,并提供了创建DirectoryServer实例的步骤以及故障切换文件系统的管理方法。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

 

Before You Start

This cookbook describes how to install Directory Server as a data service for Sun Cluster 3.1 (or higher) on Solaris 9 or 10 systems, for SPARC, x86, and x64 platforms. You install Directory Server from native packages by using the Java ES installer.

You must be familiar with Sun Cluster and Directory Server technology in order to find this cookbook useful.
A detailed How-to guide for setting up a 2 node cluster can be found here.

In following the instructions here, you create one resource group per Directory Server instance.

The example assumes that the machines are in the example.com domain. 

To Prepare Sun Cluster

Start by preparing the cluster. Directory Service requires an IP address, and also disk space. Configure the disks in failover mode with affinity set to on.

Note: You execute cluster commands scrgadm and scswitch only on one node of the cluster.
  1. /etc/hosts and /etc/nsswitch.conf
    • Make sure the logical host name you intend to use is in /etc/hosts (in this example: sun-ldap).
    • Make sure you have "files" before "nis" or "dns" in /etc/nsswitch.conf:
      hosts:      cluster files nis [NOTFOUND=return]
  2. Disks
    • Make sure the shared disks that used for the Directory Server instance do not have the global option set. Use the "no logging" option, rather than the "yes global,logging" in /etc/vfstab:
      /dev/md/sc1/dsk/d50 /dev/md/sc1/rdsk/d50 /clusteredfs/sunds ufs 2 no logging
  3. umount the disks:
    • umount /clusterdisks/sunds
  4. Create the resource group, and the logical hostname:
    • scrgadm -a -g ds-ldap1
    • scrgadm -a -L -g ds-ldap1 -l sun-ldap
  5. Create and configure the disks (HAStoragePlus in failover mode)
    • scrgadm -a -t SUNW.HAStoragePlus
    • scrgadm -a -j disks -g ds-ldap1 -t SUNW.HAStoragePlus -x FilesystemMountPoints=/clusteredfs/sunds -x AffinityOn=TRUE
  6. Enable the resource group
    • scswitch -Z -g ds-ldap1

To Install Directory Server With the Java ES Installer

Install Directory Server packages on all nodes of the cluster in their default locations (using the default BASEDIR). Do not use the Java ES installer to create or to configure a Directory Server instance. Instead, use dsadm as described in the section " To Create a Directory Server Instance" of this cookbook.

  1. Install Directory Server on all nodes of the cluster:
    • Use JES installer
    • Install all Directory Server software, including the configuration tools.
    • Do not create an instance. Do not configure anything ("configure later" in the Java ES installer).
    • Install Cluster Agents. (This is a specific checkbox in the installer).

To Create a Directory Server Instance

Create the Directory Server instance on the failover file system. Once created, manage the instance using Sun Cluster commands. Perform this procedure on only one node of the cluster.

  1. Make sure the failover filesystem is mounted on the local node:
    • scswitch -z -g ds-ldap1 -h `uname -n`
  2. Create the Directory Server instance:
    • echo secret12 > /tmp/me/password.txt
    • chmod 700 /tmp/me/password.txt
    • dsadm create -w /tmp/me/password.txt -h sun-ldap.example.com /clusteredfs/sunds/myds
      Notes:
      • The logical hostname must be specified when creating the instance, otherwise the server will use the node name. This name is used for referrals, mostly by the replication feature.
      • If you install DS as a non-root user, you need to specify port numbers higher than 1024 and make sure you have write permissions in /global/sc1/sunds/:
        dsadm create -p 1389 -P 1636 -w /tmp/me/password.txt -h sun-ldap.example.com /clusteredfs/sunds/myds
  3. Make sure the Directory Server instance is properly working:
    • Start the Directory Server instance manually on the node1:
      dsadm start /clusteredfs/sunds/myds
    • Test with an LDAP client to connect to the Directory Server instance:
      ldapsearch -h sun-ldap.example.com -b "" -s base '(objectclass=*)'
    • Stop the Directory Server instance:
      dsadm stop /clusteredfs/sunds/myds
    • Switch the Cluster to the other node:
      scswitch -z -g ds-ldap1 -h node2
    • Start the Directory Server instance manually on the other node:
      dsadm start /clusteredfs/sunds/myds
    • Test with an LDAP client:
      ldapsearch -h sun-ldap.example.com -b "" -s base '(objectclass=*)'
    • Stop the Directory Server instance:
      dsadm stop /clusteredfs/sunds/myds
  4. Enable the newly created Directory Server instance as a Cluster resource:
    • (as root) scrgadm -a -t SUNW.ds6ldap
    • (as root) dsadm enable-service --type CLUSTER /clusteredfs/sunds/myds ds-ldap1

      Note: The previous command names the resource with a predefined format which is later used to retrieve the DS instance. As a result, any attempt to change the resource name will cause the start and stop commands to fail. Also, dashes and spaces should be avoided in the installation path of the Directory Server instances. These limitations may be removed in future versions of DS 6 and its cluster agent.

  5. Work-around bug 6478568 - Missing dependency on disks in dsadm enable-service --type CLUSTER:
    • scrgadm -c -j ds--clusteredfs-sunds-myds -y Resource_dependencies=disks

To Manage a Directory Server Instance

At this point, only root can stop and start the Directory Server instance, either with the cluster commands (scswitch -e|-n|-z) on any node of the cluster, or with the dsadm command:
  • scswitch -e -j ds--clusteredfs-sunds-myds
  • dsadm start /clusteredfs/sunds/myds
内容概要:本书《Deep Reinforcement Learning with Guaranteed Performance》探讨了基于李雅普诺夫方法的深度强化学习及其在非线性系统最优控制中的应用。书中提出了一种近似最优自适应控制方法,结合泰勒展开、神经网络、估计器设计及滑模控制思想,解决了不同场景下的跟踪控制问题。该方法不仅保证了性能指标的渐近收敛,还确保了跟踪误差的渐近收敛至零。此外,书中还涉及了执行器饱和、冗余解析等问题,并提出了新的冗余解析方法,验证了所提方法的有效性和优越性。 适合人群:研究生及以上学历的研究人员,特别是从事自适应/最优控制、机器人学和动态神经网络领域的学术界和工业界研究人员。 使用场景及目标:①研究非线性系统的最优控制问题,特别是在存在输入约束和系统动力学的情况下;②解决带有参数不确定性的线性和非线性系统的跟踪控制问题;③探索基于李雅普诺夫方法的深度强化学习在非线性系统控制中的应用;④设计和验证针对冗余机械臂的新型冗余解析方法。 其他说明:本书分为七章,每章内容相对独立,便于读者理解。书中不仅提供了理论分析,还通过实际应用(如欠驱动船舶、冗余机械臂)验证了所提方法的有效性。此外,作者鼓励读者通过仿真和实验进一步验证书中提出的理论和技术。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值