HP Serviceguard for Linux Version A.11.
Legal Notices The information in this document is subject to change without notice. Hewlett-Packard makes no warranty of any kind with regard to this manual, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. Hewlett-Packard shall not be held liable for errors contained herein or direct, indirect, special, incidental or consequential damages in connection with the furnishing, performance, or use of this material. Warranty.
Table of Contents 1 Serviceguard for Linux Version A.11.19 Release Notes.....................................................................9 Announcements....................................................................................................................9 Supported Platforms and Linux Distributions...............................................................9 New Support for SUSE SLES 11......................................................................................
What You Can Do...............................................................................................24 New Features......................................................................................................25 Serviceguard Manager and Serviceguard Clusters............................................26 Current Limitations of Serviceguard Manager..................................................26 What’s Not in this Release...................................................................
Installing HP Serviceguard for Linux.................................................................................43 Pre-Installation Checklist..............................................................................................44 Pre-Installation Checklist for Serviceguard Manager..............................................45 Post-Installation Tasks...................................................................................................
List of Figures 1-1 1-2 1-3 1-4 1-5 1-6 6 Running Cluster Before Rolling Upgrade ..................................................................55 Running Cluster with Packages Moved to Node 2 ....................................................56 Node 1 Upgraded to Red Hat EL 3.............................................................................56 Node 1 Rejoining the Cluster .....................................................................................
List of Tables 1-1 Serviceguard Manager and Serviceguard Versions....................................................
1 Serviceguard for Linux Version A.11.19 Release Notes Announcements This section announces the most important features and limitations of Serviceguard for Linux A.11.19 and A.11.19.01. For more information, see “What’s in this Release” (page 13). NOTE: These Release Notes also include information about features first introduced in A.11.18 patches; see “New Features First Introduced in Serviceguard for Linux A.11.18 Patches” (page 27). Supported Platforms and Linux Distributions Serviceguard A.11.
Product Structure Serviceguard for Linux A.11.19 is available under the following product numbers: • T8720AA — A.11.19 — license for HP Integrity Servers — Variable count license for HP Integrity servers: ◦ one license required per processor ◦ license certificate has license count equal to number of licenses ordered — Installation and support services available through product options • T8721AA — A.11.
◦ one license required per host node If Serviceguard for Linux is running in virtual machines, only one license is required, irrespective of the number of virtual machines running on the host that has the license ◦ license certificate has license count equal to number of licenses ordered — one year of 24x7 support services included — Do not order this part number through a reseller When ordering through a reseller, use part number 519354–B21 • 519353-B21 — A.11.
There is also a one-time effect on rolling upgrade to Serviceguard A.11.19. Upgrade will trigger a cluster membership transition from the old to the new cluster manager when the last node in the cluster has rolled to A.11.19. This transition can take up to one second, during which time the old cluster manager will shut down and the new cluster manager will start.
Quorum Server Upgrade Required if You Are Using an Alternate Address If you are using an Alternate Quorum Server Subnet (page 32), you must upgrade the Quorum Server to version A.04.00 before you upgrade the cluster to Serviceguard A.11.19. CAUTION: If you fail to do this, the upgraded cluster will be running without a cluster lock until you have upgraded the Quorum Server. Serviceguard Manager Available from the System Management Homepage (SMH) For details, see “About Serviceguard Manager” (page 24).
New Features for A.11.19.01 (July 2009 Patches) To the existing support for Red Hat 5 and SUSE SLES 10, Serviceguard A.11.19.01 adds support for SUSE SLES 11. In addition, Serviceguard A.11.19.01 provides the following new capabilities on all three platforms. On Red Hat 5 and SUSE SLES 10, these capabilities require the patches listed under “July 2009 Patches” (page 9). • Improved support for IPv6, including support for IPv6–only hosts. See “New Support for IPv6” (page 15).
• • • • • • • • • Other minor new package features are listed under “Other Package Changes” (page 31). Serviceguard can now monitor subnets at the IP level, as well as the link level. See “About the IP Monitor” (page 21). You can now make configuration changes while the cluster is running, particularly in regard to the cluster lock, that were not allowed online in previous releases, or were restricted; see “New Online Cluster Configuration Capabilities” (page 21).
The default is IPv4. You can change it either by editing the cluster configuration file and setting the appropriate value for the HOSTNAME_ADDRESS_FAMILY parameter, or by using the new -a option of cmquerycl (1m). For more information, see “About Hostname Address Families: IPv4-Only, IPv6-Only, and Mixed Mode” in Chapter 4 of the latest version of Managing Serviceguard, at the address given under “Documents for This Version ” (page 36). Some restrictions apply, as follows.
IMPORTANT: You must use Quorum Server version A.04.00 or later. See the latest Quorum Server release notes for more information; you can find them at docs.hp.com under High Availability —> Quorum Server. The Quorum Server itself can be an IPv6–only system; in that case it can serve IPv6–only and mixed-mode clusters, but not IPv4–only clusters. • • If you use a Quorum Server, and the Quorum Server is on a different subnet from cluster, you must use an IPv6-capable router.
::1 localhost ipv6-localhost ipv6-loopback For more information and recommendations about hostname resolution, see “Configuring Name Resolution” in Chapter 5 of Managing Serviceguard. • You must use $SGCONF/cmclnodelist, not ~/.rhosts or /etc/hosts.equiv, to provide root access to an unconfigured node. See “Allowing Root Access to an Unconfigured Node” in Chapter 5 of Managing Serviceguard for more information.
IMPORTANT: Quorum Server Version A.04.00 is required if you are configuring a second Quorum Server address (QS_ADDR). This is true whether QS_ADDR is an IPv4 or an IPv6 address. Package Maintenance Mode NOTE: For information about legacy and modular packages, see Chapter 6 of Managing Serviceguard. With the patches listed under “July 2009 Patches” (page 9) Serviceguard A.11.19 allows you to perform maintenance on a modular, failover package while the package is running. This is called maintenance mode.
VLAN Configurations Serviceguard for Linux A.11.19, with the patches listed under “July 2009 Patches” (page 9), provides support for Virtual LAN configuration (VLAN). VLAN allows logical grouping of network nodes, regardless of their physical locations. You can use VLAN interfaces for both heartbeat and data networks. Configuration Restrictions Linux allows up to 1024 VLANs to be created from a physical NIC port.
which you can find at docs.hp.com under High Availability —> Serviceguard. About Package Weights Package weights and node capacities allow you to restrict the number of packages that can run concurrently on a given node, or, alternatively, to limit the total package “weight” (in terms of resource consumption) that a node can bear. For example, suppose you have a two-node cluster consisting of a large system and a smaller system.
7 of the latest version of Managing Serviceguard for Linux, and in particular the subsection “Allowable Package States During Reconfiguration”. For information about legacy and modular packages, see Chapter 6 of the latest version of Managing Serviceguard for Linux.
NOTE: PR imposes an important restriction on the use of VMware virtual machines as cluster nodes; see “Support for HPVM, VMware ESX Server, and Xen Virtual Machines” (page 32). For more information, see “About Persistent Reservations” in Chapter 3 of the latest version of Managing Serviceguard for Linux. Other Package Changes NOTE: For information about legacy and modular packages, see Chapter 6 of Managing Serviceguard. Serviceguard A.11.
About Serviceguard Manager HP Serviceguard Manager B.02.00 is a web-based, HP System Management Homepage (HP SMH) tool that replaces the functionality of the earlier Serviceguard management tools. HP Serviceguard Manager allows you to monitor, administer and configure a Serviceguard A.11.19 cluster from any system with a supported web browser. The HP Serviceguard Manager Main Page provides you with a summary of the health of the cluster including the status of each node and its packages.
New Features HP Serviceguard Manager version B.02.00 supports Serviceguard A.11.19 Red Hat RHEL 5.2 and SUSE SLES 10 and 11. The following are new capabilities in B.02.00: • Enhanced cluster monitoring: — Summarize package status iconically with drill-down to specific problem descriptions — Identify packages that cannot fail over — Identify whether nodes and packages are halted or failed — Identify package type (e.g.
Serviceguard Manager and Serviceguard Clusters If you manage clusters running different versions of Serviceguard, you will need run the version of Serviceguard Manager appropriate to each. Table 1-1 Serviceguard Manager and Serviceguard Versions Number of Clusters Serviceguard Version Serviceguard Manager Version 1 A.11.19 B.02.00 A.11.18 B.01.01 A.11.18 and later Multiple browser sessions of B.0x.
NOTE: Although Serviceguard Manager appears to support putting a package in maintenance mode, the current version will generate an error if you try to do it. • Cross-subnet configuration (see “About Cross-Subnet Configurations” (page 29)). You cannot use Serviceguard Manager to configure a cross-subnet cluster or package. What’s Not in this Release • Serviceguard A.11.19 does not support Red Hat 4. The last version that did was A.11.18. Serviceguard A.11.19 does support Red Hat 5 and SUSE SLES 10 and 11.
• A quorum server can now support up to 150 clusters comprising up to 300 nodes. The previous limits were 50 clusters and 100 nodes. The new limits are supported in Quorum Server versions A.02.00.04 and later. • In addition to the new Serviceguard demons introduced in A.11.19 (see above), new demons were introduced in a patch to A.11.
IMPORTANT: If you use DM-MPIO for the lock LUN on one node in a cluster, it must be used on all nodes. The device names should be in the form /dev/mapper/mpathp. can be different on different nodes, but must be the same; for example: • • On node1: /dev/mapper/mpath0p1 On node2: /dev/mapper/mpath1p1 For the latest information on supported configurations, see the Serviceguard for Linux Certification Matrix at the address given under “Documents for This Version ” (page 36).
• • • A minimum of two heartbeat paths must be configured for each cluster node. There must be less than 200 milliseconds of latency in the heartbeat network. Each heartbeat subnet on each node must be physically routed separately to the heartbeat subnet on another node; that is, each heartbeat path must be physically separate: — The heartbeats must be statically routed; static route entries must be configured on each node to route the heartbeats through different paths.
IMPORTANT: Although cross-subnet topology can be implemented on a single site, it is most commonly used by extended-distance clusters. For more information about such clusters, see the latest edition of HP Serviceguard Extended Distance Cluster for Linux Deployment Guide. Migrating Legacy to Modular Packages A new Serviceguard command, cmmigratepkg, first introduced in a patch to A.11.18, automates the process of migrating legacy packages to modular packages as far as possible.
indicating thatthe package is halting abnormally (for example because of the failure of a service it depends on). • • • cmviewcl -v -f line has a new field, last_halt_failed, that shows whether the last invocation of the halt script of a package on a node succeeded or failed. The value is no if the halt script ran successfully, or was not run since the node joined the cluster, or was not run since the package was configured to run on the node; otherwise it is yes.
• increases. See “About HPVM Virtual Machines and Cluster Re-formation Time” (page 33). VMware nodes. In the case of VMware nodes, Persistent Reservations (PR) handle the I/O issues. This entails the following restriction: — Serviceguard does not support configurations with more than one VMware virtual-machine node on the same host (“cluster-in-a-box”) because such configurations are not supported by PR.
IMPORTANT: This represents a net addition to the time it takes for the cluster to re-form. For example, if the cluster typically took 40 seconds to re-form before any HPVM nodes were added, it will generally take about 80 seconds when one or more VM nodes are members of the cluster, if all those nodes have the hpvminfo software. If any HPVM node without that software is a member of the cluster, it will take about 110 seconds.
For more information about Access Control Policies, see Chapter 5 of the the Managing Serviceguard for Linux manual (docs.hp.com -> High Availability), the Serviceguard Manager Help (under Defining Cluster Roles), and the cluster and package configuration files themselves. Considerations when Upgrading Serviceguard • .rhosts If you relied on .rhosts for access in the previous version of the cluster, you must now configure Access Control Policies for the cluster users.
Documents for This Version The following documents relate to Serviceguard A.11.19 and related high availability products. The latest versions can be found on docs.hp.com, under High Availability -> Serviceguard for Linux unless otherwise stated. • • • • • • • • • • • • • • Managing HP Serviceguard for Linux, Tenth Edition. This manual (sometimes referred to simply as Managing Serviceguard) has been revised for the A.11.19 release. HP Serviceguard Quorum Server Release Notes Version A.04.
http://us-support.external.hp.com (Americas and Asia Pacific) http://europe-support.external.hp.com (Europe) Compatibility Information and Installation Requirements Read this entire document and any other Release Notes you may have before you begin an installation. NOTE: All nodes must be on the same distribution, including errata level and architecture version. Using different distribution versions within the same cluster is supported only during rolling upgrade.
• • wbem-http TCP/5988 wbem-https TCP/5989 If you will be using the Quorum Server • hacl-qs 1238/TCP HA Quorum Server If you will be using the appserver utility: • hacl-poll 5315/TCP Ports Needed on SUSE: • ident 113/TCP used by identd • hacl-hb 5300/TCP High Availability (HA) Cluster heartbeat • hacl-hb 5300/UDP High Availability (HA) Cluster heartbeat • hacl-cfg 5302/TCP HA Cluster TCP configuration • hacl-cfg 5302/UDP HA Cluster UDP configuration If you will be using SNMP: • snmp 161/UDP • snmptrap 1
This port is configurable; if port 1775 is already being used by another application, configure and open another free port when you configure the firewall. System Firewalls When using a system firewall with Serviceguard for Linux, you must leave open the ports listed above. For detailed instructions, see Configuring Firewall Rules for Serviceguard on SUSE SLES 10 and Red Hat 5 on docs.hp.com under High Availability -> Serviceguard for Linux —> White Papers.
All nodes in the cluster must allow the following communications: • from the remote nodes: — TCP on ports 5302 — and allow only packets with the SYN flag — UDP on port 5302 • to the remote nodes: — TCP and UDP on dynamic ports The remote nodes must allow the following communications: • from the cluster nodes — TCP and UDP on dynamic ports • to the cluster nodes — TCP on ports 5302 — and allow only packets with the SYN flag — UDP on port 5302 Authentication communication must allow the following ports:
5. Finish the registration process. At this point you have a license key (an alphanumeric string), usually contained in a file that HP emails to you. It may also be possible to copy and paste the key directly from your browser window; in this case you will need to save the key in a file that you create. IMPORTANT: In either case, make sure you save the file and make a note of its path. (The name does not matter. In the Example (page 42) we have called it /mystuff/mySGlicense.
Example 1. 2. 3. mv $SGCONF/AutoPass/LicFile.txt $SGCONF/AutoPass/LicFile.txt.sav cp /mystuff/mySGlicense $SGCONF/AutoPass/LicFile.txt cmcheckconf –v –C where is the configuration file used to create the existing cluster. 4. • If the license is valid, keep the new license file $SGCONF/AutoPass/ LicFile.txt You can delete the old license file $SGCONF/AutoPass/LicFile.txt.sav. • If the license is invalid, restore the original license: mv $SGCONF/AutoPass/LicFile.txt.
NOTE: /etc/cmcluster.conf contains the mappings that resolve the symbolic references to$SGCONF, $SGROOT, etc., used in these Release Notes. See “Understanding the Location of Serviceguard Files” in Chapter 5 of the Managing HP Serviceguard for Linux manual for details. For more information about configuring security for prospective cluster nodes, as well as for an existing cluster, see Chapter 5 of Managing HP Serviceguard for Linux.
Pre-Installation Checklist • • Before you begin installing Serviceguard, make sure that all components of the cluster have been upgraded to their latest firmware versions. Serviceguard for Linux depends on the xinetd service. Make sure that the xinetd rpm has been installed from the distribution source (for example, your Linux installation CD) and is enabled.
◦ ◦ kernel-source scsi — For SLES 11: ◦ xinetd ◦ gcc ◦ libn1 ◦ pidentd ◦ glibc ◦ glibc-locale ◦ glibc-devel ◦ glibc-info ◦ linux-kernel-headers ◦ cpp ◦ kernel-source ◦ gcc43 ◦ sg3_utils — For Red Hat 5: ◦ xinetd ◦ gcc ◦ libn1 ◦ net-snmp ◦ lm_sensors ◦ tog-pegasus ◦ kernel-devel ◦ libgomp ◦ glibc-devel ◦ glibc-headers ◦ kernel-headers ◦ compat-libstdc++-296 ◦ compat-libstdc++-33 ◦ sg3_utils-libs ◦ sg3_utils Pre-Installation Checklist for Serviceguard Manager Before installing Serviceguard Manager make sure
NOTE: libXp is usually already installed as part of SLES 10 and 11. For Red Hat 5, you will usually need to install it explicitly from the distribution CD. 3. Downloaded and installed the Java SE Development kit (JDK): 1. Go to http://java.sun.com/javase/downloads/index.jsp NOTE: The above URL was correct at the time of publication, but may have changed in the meantime. 2. Select the Java JDK 1.6 Update 13 or later and follow the instructions for downloading the Linux rpm.
Post-Installation Tasks for Serviceguard Manager When you have finished installing the Serviceguard Manager software, do the following. 1. Restart SMH to integrate SMH with Serviceguard Manager: /etc/init.d/hpsmhd restart 2. To verify the installation, open a web browser and navigate to http:// [hostname]:2301/ The System Management Home Page appears. Go to the SMH Tools and select Serviceguard Manager.
Requirements CAUTION: • Special considerations apply to a rolling or non-rolling upgrade to Serviceguard A.11.19. See “Special Considerations for Upgrade to Serviceguard A.11.19” (page 49) • If you are using a Quorum Server, make sure you read the latest version of the HP Serviceguard Quorum Server A.04.00 Release Notes before you proceed. These are also posted on posted on docs.hp.
NOTE: HP recommends you use the rolling upgrade process if possible, because it keeps your mission-critical applications running; and also that you update the software (rather than do a cold install) if possible, because updating preserves the current OS and cluster configuration. But make sure you read the “Special Considerations for Upgrade to Serviceguard A.11.
and finished upgrading the node to the new cluster manager (Cluster Membership Protocol version 2), and then, once all nodes have been upgraded, you will see a message indicating that the new cluster has formed. Watch for three messages similar to the following (for clarity, intervening messages have been replaced with ellipses [...]): Nov 14 13:52:46 bbq1 cmcld[20319]: Starting to upgrade this node to Cluster Membership Protocol version 2 [....
• • You cannot delete Serviceguard software (via rpm -e) from a node while the cluster is in the process of rolling upgrade. This procedure depends on the upgrade or re-install keeping the same device naming convention and general system configuration. It is possible for devices to change names or be changed in the scan order in a way that cannot be corrected. If this happens, the cluster will need to be re-created rather than to be upgraded.
4. Back up the following files on media that can be easily recovered by the node after its upgrade or a new OS installation: • Host files: /root/.rhosts, /etc/hosts, /etc/profile,and the network information (including the bonding configurations): — Red Hat: /etc/sysconfig/network-scripts/ifcfg* — SUSE: /etc/sysconfig/network/ifcfg* 5.
6. Install the new Serviceguard release from the CD. The examples that follow show only the Serviceguard rpm. Generally you also need to install: • sgcmom • cmsnmpd • sgproviders(available for Red Hat 5 and SLES 10 only) To upgrade to Serviceguard Manager B.02.00, you also need to install: • • sgmgrpi hpdsau Example (Serviceguard rpm for Red Hat 5 x86_64): rpm -i RedHat5/Serviceguard/x86_64/serviceguard-A.11.19.01-0.rhel5.x86_64.
The pidentd (sg_pidentd for Red Hat 5) driver is located at: /usr/src/pidentd-/driver where version represents the version number for the pidentd that was last installed. Follow the instructions in the README file in the directory of each driver. If you have installed a new OS version, you must run a convert program. This will convert the binary file (cmclconfig) to its new release format. To run the program on the upgraded node, enter:$SGGSBIN/convert 7. 8. 9. Reboot the node.
NOTE: This and the following figures are from a use case in which the starting point of the upgrade was Serviceguard A.11.14.02 and Red Hat AS 2.1, and the rolling upgrade was to Serviceguard A.11.15.02 and Red Hat EL3. This is for illustration only; the upgrade process has not changed since. Figure 1-1 Running Cluster Before Rolling Upgrade Node1 Pkg1 SG A 11.14.02 Red Hat AS 2.1 Node2 Pkg2 SG A.11.14.02 Red Hat AS 2.1 Step 1.
Figure 1-2 Running Cluster with Packages Moved to Node 2 Node1 SG Node2 SG A.11.14.02 Pkg1 A.11.14.02 AS 2.1 Pkg2 AS 2.1 Step 2. Upgrade First Node Upgrade node1 to the new operating system release (in this example, it was Red Hat EL 3), and install the new version of Serviceguard, as shown in Figure 1-3. Figure 1-3 Node 1 Upgraded to Red Hat EL 3 Node1 Node2 Pkg1 Red Hat EL 3 Pkg2 SG A.11.14.02 Red Hat AS 2.1 Step 3.
At this point, different versions of the Serviceguard daemon (cmcld) are running on the two nodes, as shown in Figure 1-4. Figure 1-4 Node 1 Rejoining the Cluster Step 4. Repeat the Process on Next Node Repeat the process on node2. Halt the node as follows: # cmhaltnode -f node2 This causes both packages to move to node1. Then upgrade node2 to the new Linux distribution and the new version of Serviceguard. Figure 1-5 Running Cluster with Packages Moved to Node 1 Node1 Pkg1 Pkg2 Node2 SG SG A.11.15.
Step 5. Move Package Back to Original Node Move pkg2 back to its original node. Use the following commands: # cmhaltpkg pkg2 # cmrunpkg -n node2 pkg2 # cmmodpkg -e pkg2 The cmmodpkg command re-enables switching of the package, which is disabled by the cmhaltpkg command. The final running cluster is shown in Figure 1-6. Figure 1-6 Running Cluster After Upgrades Node1 Pkg1 Node2 SG A.11.15.02 Pkg2 Red Hat EL 3 SG A.11.15.
Patches for this Version Some of the capabilities described in these release notes require the “July 2009 Patches” (page 9) for Red Hat 5 and SUSE SLES 10. Before installing Serviceguard, you should also check the Hewlett-Packard IT Resource Center web page for any new patch requirements: http://itrc.hp.com (Americas and Asia Pacific) http://europe.itrc.hp.com (Europe) NOTE: Patches can be superseded or withdrawn at any time. Be sure to check the status of any patch before downloading it.
• • • • • • • • • • • • • • • • • • QXCR1000590767 (JAGag46475): cmquerycl -f line output may display incorrect VG disk device files QXCR1000589971 (JAGag45533): cmviewconf does not display package service failfast flag correctly QXCR1000740150: cmcld abort when deleting node online QXCR1000741571: hang on one system, causes complete cluster to go down QXCR1000747462: Serviceguard 11.
Problems Fixed in this Version of Serviceguard Manager The following defects have been fixed in Version B.02.00 of Serviceguard Manager.
• • • • • • • • QXCR1000790716 — Disk problems may prevent cluster creation QXCR1000840129 — NPE in ClusterConfigPage with full network probing (-w full) QXCR1000843137 — Drop-down boxes should go to the back of a tooltip QXCR1000857159 — Up/down arrows used to change sorting are sometimes hidden QXCR1000876220 — The operation log browser window may refresh itself indefinitely QXCR1000877581 — Unresponsive pull-down menu selection using Firefox 3.
