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......................................................................................
DSAU Integration...............................................................................................24 Native Language Support..................................................................................24 What You Can Do...............................................................................................24 New Serviceguard Manager Features................................................................25 Serviceguard Manager and Serviceguard Clusters...............................
Preventing Boot-Time vgscan and Ensuring Serviceguard Volume Groups Are Deactivated...............................................................................................................44 Installing HP Serviceguard for Linux.................................................................................44 Pre-Installation Checklist..............................................................................................45 Pre-Installation Checklist for Serviceguard Manager.....................
List of Figures 1-1 1-2 1-3 1-4 1-5 1-6 6 Running Cluster Before Rolling Upgrade ..................................................................56 Running Cluster with Packages Moved to Node 2 ....................................................57 Node 1 Upgraded to Red Hat EL 3.............................................................................57 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 Serviceguard version A.11.19 with the July 2009 Patches (page 9) is formally known as A.11.19.01. The October 2009 Patches (page 10) add new Serviceguard Manager features to A.11.19.01 IMPORTANT: A.11.19.01 is a prerequisite for the October 2009 patches; this means that you must apply the appropriate July patch before applying an October patch. This section announces the most important features and limitations of Serviceguard for Linux A.
• • • SGLX_00267 for Serviceguard A.11.19 on SUSE SLES 10 IA32 SGLX_00268 for Serviceguard A.11.19 on SUSE SLES 10 IA64 SGLX_00269 for Serviceguard A.11.19 on SUSE SLES 10 x86–64 October 2009 Patches See “New Features for October 2009 Patches” (page 14) for information about new features introduced in the following patches: • SGLX_00270 for Serviceguard A.11.19.01 on Red Hat • SGLX_00273 for Serviceguard A.11.19.01 on SUSE IMPORTANT: A.11.19.
• 519354-B21 — A.11.19 — license for HP ProLiant servers — One license for HP ProLiant server ◦ 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 — one year of 24x7 support services included • 519355-B21 — A.11.
IMPORTANT: Do not delete this file. New Cluster Manager Serviceguard A.11.19 introduces a new cluster manager. In a running cluster, this affects node timeout and failover and cluster re-formation, providing significant performance improvements; see the discussion of MEMBER_TIMEOUT under “Cluster Configuration Parameters” in the latest version of Managing Serviceguard for Linux for more information. There is also a one-time effect on rolling upgrade to Serviceguard A.11.19.
Quorum Server Upgrade Required if You Are Using an Alternate Address If you are using an Alternate Quorum Server Subnet (page 33), 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).
NOTE: As of the date of these Release Notes, there is no support for the WBEMProvider on SLES 11. If you need this functionality, email SGLXhelp@hp.com to get the latest status. 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.
IMPORTANT: A.11.19.01 is a prerequisite for the October 2009 patches, and hence for the new Serviceguard Manager features. This means that you must apply the appropriate July patch (see“July 2009 Patches” (page 9)) before applying an October patch. Features Introduced in A.11.19 The features being introduced in the July and October 2009 patches are listed in the previous subsections; see New Features for A.11.19.01 (July 2009 Patches) and “New Features for October 2009 Patches” (page 14).
• • You can now configure a maximum of 300 packages in a cluster. The previous limit was 150. There are new Serviceguard demons: — Service Assistant Daemon: cmserviced (replaces cmsrvassistd) — Cluster WBEM Agent Daemon: cmwbemd — Proxy Daemon: cmproxyd See the section “Serviceguard Daemons” in Chapter 3 of the latest edition Managing Serviceguard for Linux for more information about these daemons. About the New Features New Support for IPv6 Serviceguard A.11.
::1 localhost ipv6-localhost ipv6-loopback • • All addresses must be IPv6, apart from the node's IPv4 loopback address, which cannot be removed from /etc/hosts. The node's public LAN address (by which it is known to the outside world) must be the last address listed in /etc/hosts. Otherwise there is a possibility of the address being used even when it is not configured into the cluster. • You must use $SGCONF/cmclnodelist, not ~/.rhosts or /etc/hosts.equiv, to provide root access to an unconfigured node.
Recommendation for IPv6-Only Mode If you decide to migrate the cluster to IPv6-only mode, you should plan to do so while the cluster is down. Rules and Restrictions for Mixed Mode • Red Hat 5 clusters are not supported. NOTE: This also applies if HOSTNAME_ADDRESS_FAMILY is set to IPv6; Red Hat 5 supports only IPv4-only clusters.
In addition, you cannot use cmquerycl -q to configure IPv6 addresses (or hostnames that resolve only to IPv6 addresses) for QS_HOST and QS_ADDR. You must generate a cluster template file, edit it to set the appropriate (IPv6) values for QS_HOST and QS_ADDR and set HOSTNAME_ADDRESS_FAMILY to ANY, and then use cmapplyconf to apply the configuration. See the latest version of the HP Serviceguard Quorum Server Version A.04.00 Release Notes on docs.hp.
For more information, rules, and instructions see “Maintaining a Package” in Chapter 7 of the latest version of Managing Serviceguard, which you can find at the address given under “Documents for This Version ” (page 37). 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.
For more information, see “About Package Dependencies” in Chapter 4 of the latest version of Managing Serviceguard for Linux, and in particular the subsection “Extended Dependencies”. See also the white paper Serviceguard’s Package Dependency Feature, which you can find at docs.hp.com under High Availability —> Serviceguard.
New Online Package Configuration Capabilities Serviceguard A.11.19 allows you to perform most configuration changes to modular packages, and many changes to legacy packages, while the package is running. For details, instructions, and caveats, see the section “Reconfiguring a Package” in chapter 7 of the latest version of Managing Serviceguard for Linux, and in particular the subsection “Allowable Package States During Reconfiguration”.
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 33). 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 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 Serviceguard Manager Features HP Serviceguard Manager supports Serviceguard A.11.19, Red Hat RHEL 5.2, and SUSE SLES 10 and SLES 11. The following are new capabilities: • 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.
New Serviceguard Manager Features for October 2009 Patches With the patches listed under “October 2009 Patches” (page 10), Serviceguard Manager provides the following new capabilities: • Cluster Topology Map. Coordinates all elements of the HA monitoring infrastructure into a single, easy-to-use graphical format.
TIP: To prevent an OutofMemory error reported by Tomcat (Exception in thread "main" java.lang.OutOfMemoryError: Java heap space), which may occur especially if the server is under heavy load or Serviceguard Manager is managing a large cluster (16 nodes with 300 packages), do the following from the command line: 1. Stop hpsmhd /etc/init.d/hpsmhd -stop 2. Modify /opt/hp/hpsmh/tomcat/bin/startup.sh , adding the following line in the export statements section: export CATALINA_OPTS="-Xms512m -Xmx512m" 3.
Features Introduced Before A.11.19 New Features First Introduced in Serviceguard for Linux A.11.18 Patches The following features were first introduced in patches to A.11.18: • • • • • • • Serviceguard supports DM-MPIO; see “Support for DM-MPIO” (page 29). Serviceguard supports cross-subnet configurations. See “About Cross-Subnet Configurations” (page 29). Serviceguard provides a means of automatically migrating existing (legacy) packages to modular packages.
• • A new command, cmdisklock, can be used to repair or replace a lock LUN; see “Replacing a Lock LUN” in Chapter 8 of Managing Serviceguard for Linux, and the cmdisklock manpage, for more information. Serviceguard for Linux version A.11.18 introduced initial support for two application interfaces based on the Service Availability Forum Specification (SAF, saforum.org). These are for use only by ISVs (Independent Software Vendors) with special support contracts.
Configuration Tasks Cluster and package configuration tasks are affected as follows: • • You must use the -w full option of cmquerycl to discover actual or potential nodes and subnets across routers.
monitored for this package are configured for PARTIAL access, each node on the node_name list must have at least one of these subnets configured. — As in other configurations, a package will not start on a node unless the subnets configured on that node, and specified in the package configuration file as monitored subnets, are up. • • Cross-subnet configurations are not supported in IPv6-only or mixed mode clusters (see “New Support for IPv6” (page 16) for information about these modes).
NOTE: The cmmigratepkg command requires Perl version 5.8.5 or higher on the system on which you run the command. It should already be on the system as part of the Linux distribution. Other Package Changes The patches listed under “New Features First Introduced in Serviceguard for Linux A.11.18 Patches” (page 28) provide the following new capabilities: • Serviceguard now supplies information explaining why a package has shut down.
cmgetconf -c clusterA -P /tmp/pkgconfig Alternate Quorum Server Subnet Serviceguard 11.19 allows you to configure an alternate subnet for communication between the cluster nodes and the Quorum Server. You can do this from the command line or in Serviceguard Manager. For details and instructions, see the HP Serviceguard Quorum Server Version A.04.00 Release Notes at http://docs.hp.com -> High Availability -> Quorum Server. IMPORTANT: As of Serviceguard A.11.
• increases. See “About HPVM Virtual Machines and Cluster Re-formation Time” (page 34). 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 43) 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 50) • 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 58 Node2 SG SG A.11.
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). 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.
• • • • QXCR1000924513: cmcld SIGSEGV when starting a 1 node cluster after upgrade to A.11.19 QXCR1000939674: cmcld abort due to closing the same fd twice after QS failure QXCR1000938745: Serviceguard 11.19 cmproxyd reports cluster is not configured when it is QXCR1000941884: Serviceguard 11.19 unlimited modular package service restarts is impossible Problems Fixed in this Version of Serviceguard Manager The following defects have been fixed in Version B.02.00 of Serviceguard Manager.
Defects • • • QXCR1000814148: Unable to connect to QS at IP address and port 1238 QXCR1000882834: Package halt behavior for modular packages. QXCR1000939110: ServiceGuard commands timeout due to delayed response from cmclconfd Known Problems for Serviceguard Manager The following are known problems for Serviceguard Manager B.02.
In addition, the following documents are available in Japanese language editions, which are included on the HP Serviceguard for Linux CD: • 64 Managing HP Serviceguard for Linux, Ninth Edition Serviceguard for Linux Version A.11.